Another quick experiment in harpsichord overtone perception

Primarily for Brad, but anyone else is of course welcome to join in.

Let's test, in a verifiable manner, the ability to detect the 7th
partial in the sound of a real harpsichord:

http://freesound.iua.upf.edu/packsViewSingle.php?id=1491

Granted, it is not an Italian, but that is not important. Nor is the
quality of the recording (which is pretty damn good, actually), nor
the distance to the strings, nor any of that. It's just about what
each person can hear in a verifiable and objective sound sample of a
real instrument. No excuses, Brad, just tell us if you think the 7th
harmonic is present, and if so, at what level, say perhaps over a
range of three possibilities, which we could say would be:

1. weak, hardly audible-2. normal, similar to other readily heard
partials, especially those one might use to adjust commonly tempered
intervals (fifths, thirds, etc)
3. strong, it sticks out like a sore thumb

Additionally, for each note, you could tell us which partials you can
hear quite clearly. No need to rate their relative strengths, just
which ones you easily hear.

CAUTION: DO NOT use external speakers of any kind! Use only a
reasonably good set of headphones. This is a standard requirement for
ANY test for beating or timbre evaluation using recorded audio
examples!! The use of speakers can invalidate the results, because
your head position relative to them can cause phase cancellation which
can eliminate content at a specific frequency even though it may
actually be present in the sound. With headphones, this cannot happen,
unless of course the sample has been badly mixed, that is, in post
production the sound of two mics has been combined into one common
channel.

After Brad has given us his answers, I will post spectrograms of each
note, so that we can compare the reality to Brad's capability to use
his ears as a reliable FFT device.

If you want, you can ask your friend the violinist/former organ voicer
to do the same, and give us his idea as to the spectral content.

However, in the interest of science, and as a learning experience for
you and anybody else who cares to try it, please, nobody cheat and
look at a real spectrogram as you are listening, OK?

Don't waste your time going any higher than the tenor b example (file
028), as I can tell you already, there is almost nothing there except
the fundamental and the 2nd partial. However, if someone really wants,
go ahead and listen to higher notes, and I'll do them as well.

Happy Trails!

P

Upon careful scrutiny, I sensed the 7th harmonic starting from number 11. It
is very pronounced. 15 is faint, 19 is fainter, 24 is brigher, 28 is
detectable, 32 is same, 36 is diminished, above this is a sudden drop.

Oz.

----- Original Message -----
From: "Paul Poletti" <paul@polettipiano.com>
To: <tuning@yahoogroups.com>
Sent: 21 �ubat 2008 Per�embe 10:39
Subject: [tuning] Another quick experiment in harpsichord overtone
perception

> Primarily for Brad, but anyone else is of course welcome to join in.
>
> Let's test, in a verifiable manner, the ability to detect the 7th
> partial in the sound of a real harpsichord:
>
> http://freesound.iua.upf.edu/packsViewSingle.php?id=1491
>
> Granted, it is not an Italian, but that is not important. Nor is the
> quality of the recording (which is pretty damn good, actually), nor
> the distance to the strings, nor any of that. It's just about what
> each person can hear in a verifiable and objective sound sample of a
> real instrument. No excuses, Brad, just tell us if you think the 7th
> harmonic is present, and if so, at what level, say perhaps over a
> range of three possibilities, which we could say would be:
>
> 1. weak, hardly audible-2. normal, similar to other readily heard
> partials, especially those one might use to adjust commonly tempered
> intervals (fifths, thirds, etc)
> 3. strong, it sticks out like a sore thumb
>
> Additionally, for each note, you could tell us which partials you can
> hear quite clearly. No need to rate their relative strengths, just
> which ones you easily hear.
>
> CAUTION: DO NOT use external speakers of any kind! Use only a
> reasonably good set of headphones. This is a standard requirement for
> ANY test for beating or timbre evaluation using recorded audio
> examples!! The use of speakers can invalidate the results, because
> your head position relative to them can cause phase cancellation which
> can eliminate content at a specific frequency even though it may
> actually be present in the sound. With headphones, this cannot happen,
> unless of course the sample has been badly mixed, that is, in post
> production the sound of two mics has been combined into one common
> channel.
>
> After Brad has given us his answers, I will post spectrograms of each
> note, so that we can compare the reality to Brad's capability to use
> his ears as a reliable FFT device.
>
> If you want, you can ask your friend the violinist/former organ voicer
> to do the same, and give us his idea as to the spectral content.
>
> However, in the interest of science, and as a learning experience for
> you and anybody else who cares to try it, please, nobody cheat and
> look at a real spectrogram as you are listening, OK?
>
> Don't waste your time going any higher than the tenor b example (file
> 028), as I can tell you already, there is almost nothing there except
> the fundamental and the 2nd partial. However, if someone really wants,
> go ahead and listen to higher notes, and I'll do them as well.
>
> Happy Trails!
>
> P
>
>
>
>
>
> You can configure your subscription by sending an empty email to one
> of these addresses (from the address at which you receive the list):
> tuning-subscribe@yahoogroups.com - join the tuning group.
> tuning-unsubscribe@yahoogroups.com - leave the group.
> tuning-nomail@yahoogroups.com - turn off mail from the group.
> tuning-digest@yahoogroups.com - set group to send daily digests.
> tuning-normal@yahoogroups.com - set group to send individual emails.
> tuning-help@yahoogroups.com - receive general help information.
>
> Yahoo! Groups Links
>
>
>

Well, I've already wasted more than an hour of non-lucrative time trying to get this set up. All so I can report my hearing (yet again), and then it'll probably get greeted with yet more rhetorical spin selected from a digital output to make me look wrong (or at least unable to hear properly to the satisfaction of a person who hasn't met me). First register for and "activate" a membership to that web site, through three rounds of e-mail delays; then download large files; then unzip them; then try to open them. I doubt my low-grade Koss headphones would be to your satisfaction either, as to flatness of response or noise cancelling, but I'll try it.

If this patronizing experiment is to continue, please condescend to recommend a free Windows Vista program that is able to play .aif files. Windows Media Player hadn't a clue. I installed an old copy of WinAmp I had sitting around, but it too doesn't recognize .aif. My old copy of MusicMatch Jukebox doesn't even open anymore. RealPlayer opens one of the .aif files, and apparently plays through it, but gives no sound (and yes, the sound card is properly enabled...it plays other things, including the sound effects of opening/closing these programs).

I'd like to try this experiment, if only to see the objective analysis of those harpsichord tones (forget the anti-Lehman spin to the rhetoric!), but until I can open the files I have more important and lucrative things to go do. Please recommend a suitable player.

Brad Lehman

Audacity will read aif on Mac, and as it is also free for M$, Linux etc, so it may well serve your purposes.

> http://audacity.sourceforge.net/download/
>

>
>
> I'd like to try this experiment, if only to see the objective analysis
> of those harpsichord tones (forget the anti-Lehman spin to the
> rhetoric!), but until I can open the files I have more important and
> lucrative things to go do. Please recommend a suitable player.
>
> Brad Lehman
>
>

>
>
>
Charles Lucy
lucy@lucytune.com

- Promoting global harmony through LucyTuning -

for information on LucyTuning go to:
http://www.lucytune.com

For LucyTuned Lullabies go to:
http://www.lullabies.co.uk

--- In tuning@yahoogroups.com, Brad Lehman <bpl@...> wrote:
>
> Well, I've already wasted more than an hour of non-lucrative time
trying
> to get this set up.

I don't understand. Can't you just listen to them directly in your web
browser by clicking on the little play triangle next to the file name?
No need to download them. The browser versions do have a touch of
distortion in the initial transients of the pluck, but after that they
are essentially identical, especially in harmonic content after the
first 1/2 second or so, which is exactly when you listen for things
like beating and spectrum.

What browser are you using? I just works in Firefox, and as far as I
know, FF is the same on Win-doze. But then Windows computers do suffer
from all sorts of sound card nonsense, so maybe that's the prob. I
know, 'cause I'm forced to use Windows at school, and it's always a
pain in the arse to do anything.

> ...(or at least unable to hear
> properly to the satisfaction of a person who hasn't met me).

You keep bringing that up, and I don't understand why. You may be a
tremendously personable chap... wouldn't change one iota about
technical or acoustical matters. Either you can hear something which
is there or you can't. Either something you claim you hear is there or
it isn't.

> First
> register for and "activate" a membership to that web site, through
three
> rounds of e-mail delays; then download large files; then unzip them;
> then try to open them.

Like I said, no need to do so.

> I doubt my low-grade Koss headphones would be to
> your satisfaction either, as to flatness of response or noise
> cancelling, but I'll try it.

They can't be much worse than my 17 euro Sony MDR-201. I just wanted
to caution people against using those super cheap ear bud-like
thingees, like the ones you can pick up for 1,95 a pair at the
fleamarket, or the ones they give you for free on RENFE high speed trains.

>
> If this patronizing experiment is to continue, please condescend to
> recommend a free Windows Vista program that is able to play .aif files.

Well, if you want to see it as a patronizing experiment, fine. Why
participate, then? I thought it was merely an opportunity for us all
to learn, as I have, that lots of times you can't always hear what's
in a spectral mix.

> Windows Media Player hadn't a clue.

Of course not. Evil Empire software only recognizes Evil Empire
formats (like this new word processing format .docx Microsuck is
trying to ram down everybody's throat so that you have to buy MS
Office in order to open a letter). It wouldn't do anything so
mindbogglingly useful as reproduce the most common audio file format
in the world, now would it?

> I installed an old copy of WinAmp
> I had sitting around, but it too doesn't recognize .aif. My old
copy of
> MusicMatch Jukebox doesn't even open anymore. RealPlayer opens one of
> the .aif files, and apparently plays through it, but gives no sound
(and
> yes, the sound card is properly enabled...it plays other things,
> including the sound effects of opening/closing these programs).
>

Oy oy oy! Buy a Mac and be done with it!

Alternatively, download iTunes for Windows (free from the apple website):

http://www.apple.com/itunes/download/

It plays just about everything, and has a pretty dandy converter
to/from MP3, Aif, and several other formats. Though it is a bit clunky
in terms of selecting conversion formats, I will admit.

> I'd like to try this experiment, if only to see the objective analysis
> of those harpsichord tones (forget the anti-Lehman spin to the
> rhetoric!), but until I can open the files I have more important and
> lucrative things to go do. Please recommend a suitable player.

Don't you ever listen to CD's on your computer? It's the same format.

Anyway, I know iTunes will do it cause I also downloaded them (just
cause I wanted them for other purposes) and that's what I use for
playback. They also play just fine in Garageband, and they play no
prob out the Mac Finder preview window as well. Haven't tried them in
my more esoteric audio players, except a really old version of Sound
Studio, which chokes on them unless I import them into iTunes, copy
the file, and then export the copy to the desktop again. Then they run
fine. Go figure.

Good luck, hope you can work it out!

Ciao,

P

--- In tuning@yahoogroups.com, "Paul Poletti" <paul@...> wrote:

> Oy oy oy! Buy a Mac and be done with it!

<OT geek talk from an ubuntu fan>
Sure, trade one 'evil empire' for another slightly-less-evil one.

Go Linux! Anyone with half a brain can download Ubuntu Linux, one of
the most technically advanced platforms ever, for free. No old-school
geek only Linux there---everything just works, pretty much. And then
you'd really be done with it. Evil empires, that is. Linux is the true
dream of socialism in action--people devoting their time, talent and
energy to make things work, just for the sake of it--information
technology for free. (Yes, there is some corporate $$$ thrown in the
pool, too)

Of course, I don't see people making hardware for free anytime soon,
so it's an interesting isolated phenomenon.

still, I admire Apple for their innovation.
</OT geek talkfrom an ubuntu fan>

Of course not! About 2/3rds of my colleagues (at least) in the
Sonology department are running Linux on their PC's. It's just that
since Brad seemed to be challenged by doing something so basic and
simple as getting an AIF(F) file to play on Windows, I thought
suggesting Linux would be a bridge too far. Funny, since according to
his on-line CV, it would seem as though his main gig in life is some
sort of software work.

Go figure...

Ciao,

P

In the first quiet computer time since Saturday, I've now set up QuickTime so the thing will at least resemble a Mac. I'll try Audacity later, and thanks for the link.

Sure enough, the files play now...after downloading and unzipping.
http://freesound.iua.upf.edu/packsViewSingle.php?id=1491
I would have to do some re-association of MIME types in the browser to get any of this to play automatically.

My comments are below.

Paul wrote:
> Granted, it is not an Italian, but that is not important. Nor is the
> quality of the recording (which is pretty damn good, actually), nor
> the distance to the strings, nor any of that. It's just about what
> each person can hear in a verifiable and objective sound sample of a
> real instrument. No excuses, Brad, just tell us if you think the 7th
> harmonic is present, and if so, at what level, say perhaps over a
> range of three possibilities, which we could say would be:
>
> 1. weak, hardly audible-2. normal, similar to other readily heard
> partials, especially those one might use to adjust commonly tempered
> intervals (fifths, thirds, etc)
> 3. strong, it sticks out like a sore thumb

Yes, its being a non-Italian IS that important. The Italian single that I played again all morning Saturday (same one as last week) has a strikingly different timbre from the samples here...while it behaves similarly to my Italian virginal here. In the temperament-setting region at the middle of the keyboard, its minor 21st (overtone 7) and the 23rd (overtone 9) are in there just barely long enough to assess a 7:4, 7:5, or 9:4 directly and zero out their beats. At anything above two or three seconds after the initial pluck, all that stuff has already died away. The sustained tone has hardly anything but roots, 5ths, and a faint major 17th.

That is to say, after the first two seconds of playing the notes on Italian-styled harpsichords as these, almost anything close to septimals (or for that matter to 45/32) sounds beatless in practice. That's even more so when listening from playing position, which is a valid (while not optimal) place to do listening; as Paul helpfully explained on the technical side, and as I described empirically last week, there's cancellation. It's a valid place to sit and listen now, and we have to consider the possibility that it was also a valid place to sit and listen for the now-long-dead theorists trying things at their own harpsichords. And to tune anything accurately expecting to get beats and use them, it's necessary to reiterate the notes several times, focusing on the first second or two of the sound.

The cancellation at the player's position of a normally-shaped harpsichord is not so much an issue on a virginal, because the raised lid is sending the sound straight to and past the player's body. But as I said at the start of these threads, which got pooh-poohed and worse, overtones 7 and up drop out so quickly that 45/32 sounds BEATLESS within a couple of seconds after the pluck. When I'm playing in 1/6 syntonic on this virginal, despite the direct sound and the theoretical presence of beating involving overtone 7, all twelve tritones (which are either 45/32 or 64/45) sound pure and beatless to me...which sound is part of the attraction of setting up 1/6 syntonic in the first place. As soon as the noisy jangle of striking two-note or bigger chords has passed, within a second or so, the instrument rings out sounding resonant and beatless in what's left. I've tested it again here this morning, with the virginal open and the room as silent as I can make it.

On Saturday at the Italian harpsichord I also made sure the HVAC system was turned off most of the time, because right there that noise eats up whatever overtones would have been audible. Not a 17th-18th century problem, of course; lucky them for not having to deal with any fans. That instrument Saturday was a more refined-sounding (i.e. less noisy immediately after pluck time) tone than I get here, and it had a little more 7 in it, but the fundamentals and the quint are so much stronger and better-sustained that 7 is negligible after two or three seconds.

Last night we had a concert with 20 students plus me as guest, playing Bach (and in my case Kuhnau) on a Steinway grand in the Bach temp. The piano tuner did a terrific job with his unisons, octaves, and the evenly circulating 1/6 Pythagorean comma 5ths. Lots of chords that should have had all kinds of beating in them, theoretically on paper, simply didn't at the audience positions; the instrument's sustained tone (pianos being weak in upper harmonics anyway...) sounded beatless at several meters away. The C major and F major triads have a 1/6 PC 5th and a 3/11 SC sharp major 3rd in them, yet they sound utterly calm. So does G major, with the same 5th but a +5/11 SC major 3rd. Hearing the students play in E major, B major, C# major in the same pieces I am accustomed to playing and listening to myself on harpsichords, but here on piano, everything was so remarkably INactive by comparison with harpsichord that sometimes it didn't even seem like those keys/scales to me. It merely sounded like a startlingly well-in-tune piano being almost beatless most of the time, whether the harmonies were simple or complex. (And OK, some of the students did muddy things up by pedalling way too much, but when their foot was off the pedal the piano sounded clean.)

I walked around to different spots in the hall as some of them rehearsed, and found the response to be pretty consistent everywhere I went: no noticeable beats in music where it would be much more active in the same music on harpsichord. The point of these remarks is that an instrument's timbre affects perception so much, and so do the distance and cancellation patterns, that any beating can be vastly different in theory vs musical practice. Who's to say what a best listening spot is? And for what purpose? Musical blend? Clearest Affekt? As part of the talk I played out a couple of short wild-key Beethoven and Chopin excerpts I had rehearsed at home on harpsichord, and on this piano they sounded so much tamer I could scarcely believe it was the same temperament. But it was, on close inspection of all the 4ths and 5ths.

Now, back to Paul's web samples that don't have an Italian-harpsichord tone. I'll listen on QuickTime with Koss TD-61 headphones; not great but the best I have available. The computer fan is quiet but not eliminated by the headphones. And, of course, the recordings of these individual notes weren't made from anywhere near playing position.... And my usual way of assessing harmonics on harpsichord is not through single notes, but through having the liberty of playing a second note simultaneously (either above or below the fixed note being assessed) and moving it myself to find beat strengths. All that said:

The game is to pick out the 7th harmonic....

- Clav002: "2, normal" - it's fairly easy to pick out the 7th here and it's well-sustained, while the 3rd and 6th harmonics (both playing the 5th degree above the root) are plenty stronger than the 7th is. The sample becomes useless at the 10-second mark where it pulses with unnatural noise. The 5th overtone (major 17th) makes its most noticeable appearance about eight seconds after the pluck.

- Clav007: "1, weak" - the 3rd harmonic is so strong in there that it's hard to hear much else, until the 5th one peeks around it after about six or seven seconds. I can't hear any useful 7th harmonic in this tone.

- Clav011: "1.5, on the weak side of normal" - this tone has so much inharmonic beating in it by itself, a conflicting resonance at the soundboard, that it's hard to pick out much...especially among the odd harmonics. The 2nd and 4th harmonics (octaves) are so strong and noticeable, past the inharmonicity, that they pretty much win. I can hear only a ghost of a 7th in there, and scarcely enough to be useful.

- Clav015: "2, normal" - a decent amount of 7 in there, and a good balance all around. The strongest harmonic in here, over most of the sustain, is 2. I can find 2, 3, 4, 5, 6, 7, and 8 in here without any problem. What's with the chirping that starts after about six seconds?

- Clav019: "1, weak" - the strong 3 and 4 in there pretty much block out my ability to hear much above them. I can catch a whiff of 5, but it comes and goes.

- Clav024: "2, normal" - a decent bit of 7, but it takes a couple of seconds to pick it out. 6 and 3 are pretty strong, and there's plenty of 5. The inharmonicity down at the fundamental is distracting; so is the birdie.

- Clav028: "1.5, weak to normal" - I can pick out 7, faintly. 2, 3, and 4 are strong and well balanced. The birdie's still having fun. Starting at nine seconds in, what's with the noise in the guy's recording equipment that I've also heard in several of the others? Is there an auto-level routine in there trying to compensate for the decrescendo, that late in each tone?

- Clav032: "1, weak" - I'm not hearing much beyond 1, 2, and 4 after the pluck has died off. It's hard to find 3 in this one. The garbage in there after nine seconds is tiresome.

- Clav036: "1, weak" - after the first 0.25 second there's just not much to hear in there anymore but 1, 2, and a whiff of 4. At nine seconds, it's not only the chuffing thing but also a pitched beeping!

- Clav040: "1, weak" - this one has a little more 3 in it than the preceding, but again after the pluck is gone there's not much to grab but 2 and a little bit of 4. There's enough of 5 in there for about one second that I could probably use it to zero a major 3rd or 10th...but it's getting so faint and transient that I'd be playing the notes repeatedly and often to get it.

- Clav044: "1, weak" - after two seconds there's hardly anything left in there but 1 and 2. Before that I'm hearing some inharmonicity, perhaps a conflict between 2 and 4. 5, 6, and 7 are so negligible in this tone as to be useless. Not much 3, either.

- Clav048: "1, weak" - 4 is there only briefly, then it goes away and I hear 2 for about one more second, and then after that there's not much but 1. Ping. The chuffing at the end reminds me of the movie "Contact" where Jodie Foster's character picks up a signal from outer space.

- Clav052: "1, weak" - hardly anything but 1 and 2, and then 2 goes away.

- Clav056: "1, weak" - 1 and 2 have inharmonicity, or maybe it's 1 with the soundboard. I hear a whiff of 3, briefly. And oddly enough, there's more 5 here (briefly too) than in any of the other samples above Clav040. But the frequency of 5 is up there where I'm not going to hear much higher than it, usefully.

And now into the ether go these observations. I hope it was worth my time. At least it was interesting. Time to go tune the semi-brutal virginal from scratch (will probably do a regular 1/5), and then go do one of the Flemishes, and pick out 10 minutes of solo Couperin for a French chamber concert with colleagues.

Brad Lehman

--- In tuning@yahoogroups.com, Brad Lehman <bpl@...> wrote:

>
> Paul wrote:
> > Granted, it is not an Italian, but that is not important.

> Yes, its being a non-Italian IS that important. The Italian single
that
> I played again all morning Saturday (same one as last week) has a
> strikingly different timbre from the samples here...while it behaves
> similarly to my Italian virginal here.

Your mixing apples and oranges again. Yes, of course it is important
if we want to verify your judgment of the tone of your Italian, but
that can only be done with a recording of YOUR Italian, not any old
Italian. But that's not want we're on about. This test is simply to
evaluate your hearing, and that of anybody else who cares to try it,
for that matter. It could be any kind of instrument.

Regarding whether or not the 45/32 tritone would have been considered
anything uniquely compelling by 18th century composers, it is
completely irrelevant, because not everybody used Italians.

> In the temperament-setting
> region at the middle of the keyboard, its minor 21st (overtone 7) and
> the 23rd (overtone 9) are in there just barely long enough to assess a
> 7:4, 7:5, or 9:4 directly and zero out their beats. At anything above
> two or three seconds after the initial pluck, all that stuff has
already
> died away. [several paragraphs more of comments about the
difficulties of tuning harpsichords]

All true enough, but is it relevant? Remember Weckmeister recommends
that the student of tempering can best learn to hear using a regal,
precisely because it is immune to more or less all of these problems,
including (since the sound generators are so close to one another),
that of cancellation. So whether or not all of this is important in
regards to how 18th century folks would have thought about tempering
is questionable.

>
> Last night we had a concert with 20 students plus me as guest, playing
> Bach (and in my case Kuhnau) on a Steinway grand in the Bach
temp.[snip] As part of
> the talk I played out a couple of short wild-key Beethoven and Chopin
> excerpts I had rehearsed at home on harpsichord, and on this piano they
> sounded so much tamer I could scarcely believe it was the same
> temperament. But it was, on close inspection of all the 4ths and 5ths.

I think it is of little or no value to set historical temperaments on
modern pianos. They are so intensely inharmonic and constructed so as
to be largely overtone deficient that it has no real relevance to what
people did on brighter harmonic instruments. It's sort of like setting
meantone on a Gamelan. Interesting sounds, yeah, but so what?

> And, of course, the recordings of these
> individual notes weren't made from anywhere near playing position....

Again, the point is merely to test your ability to accurately judge
the spectral content of the sound. No excuses!

>
> The game is to pick out the 7th harmonic....
Right! Haven't had time to check them all, but I can already say some
of your comments are spot on, others are way off the mark. I'll whip
up a score later this week. Right now I have to finish preparing the
slide show for tomorrow's class session on, as fate would have it, the
acoustic basis for consonance and dissonance.

Anybody else care to try in the meantime?

Ciao,

P

> I think it is of little or no value to set historical
> temperaments on modern pianos. They are so intensely
> inharmonic and constructed so as to be largely overtone
> deficient that it has no real relevance to what people
> did on brighter harmonic instruments. It's sort of like
> setting meantone on a Gamelan. Interesting sounds, yeah,
> but so what?

It's nothing like setting meantone on gamelan. I think
I've asked you before for evidence that pianos are
constructed to lack overtones. Also, the inharmonicity
is not nearly as severe as you assert.

-Carl

--- In tuning@yahoogroups.com, "Carl Lumma" <carl@...> wrote:
>
> > I think it is of little or no value to set historical
> > temperaments on modern pianos. They are so intensely
> > inharmonic and constructed so as to be largely overtone
> > deficient that it has no real relevance to what people
> > did on brighter harmonic instruments. It's sort of like
> > setting meantone on a Gamelan. Interesting sounds, yeah,
> > but so what?
>
> It's nothing like setting meantone on gamelan. I think
> I've asked you before for evidence that pianos are
> constructed to lack overtones. Also, the inharmonicity
> is not nearly as severe as you assert.

I agree with Carl.

-AKJ.

> I think it is of little or no value to set historical temperaments on
> modern pianos. They are so intensely inharmonic and constructed so as
> to be largely overtone deficient that it has no real relevance to what
> people did on brighter harmonic instruments.

"Little or no value" in what way? The more than 100 people who showed
up last night to hear this free concert acted as if they had a good
time, being musically pleased by the solo pianism of 21 different
musicians. I wouldn't suggest that it proves anything, other than its
own ends: a musically effective way ("it works beautifully") to hear
Bach and Kuhnau selections on a Steinway grand, spending 2 1/2 hours
indoors on a wintry Sunday evening. The full-time piano technician
who set it up reported back today (both to me and to his five piano
faculty members) that he has had a great time learning this one and
hearing it, and he wants to have more discussion.

If we can encourage fine musicians and technicians to do and hear
enjoyable things, re-examining their own craft in a new and exciting
way, isn't that an important mission for anybody interested in
non-mainstream tuning schemes? Ear-stretching beauty?

As for "constructed so as to be largely overtone deficient," I thought
the mission of modern piano builders was to create instruments that
people enjoy playing and listening to, and buy many of...and they'll
use whatever tonal design helps them to achieve that goal in a
cost-effective and perhaps also artistic way. Maybe I'm mistaken in
that naive impression.

Brad Lehman

--- In tuning@yahoogroups.com, "Brad Lehman" <bpl@...> wrote:
>
> > I think it is of little or no value to set historical temperaments on
> > modern pianos. They are so intensely inharmonic and constructed so as
> > to be largely overtone deficient that it has no real relevance to what
> > people did on brighter harmonic instruments.
>
>
> "Little or no value" in what way?

In terms of what we have been talking about for weeks now, Brad:
judging temperaments for 18th century music. And besides, that is what
your whole "Bach" hypothetical temperament is all about. Or are we
going to get into one of those "If Bach had known about the modern
piano, he certainly would have put MY (er, sorry, his) temperament on
it!" discussions?

> The more than 100 people who showed
> up last night to hear this free concert acted as if they had a good
> time, being musically pleased by the solo pianism of 21 different
> musicians.

I'm sure they did. I'm sure they would have regardless of the
temperament. As you said yourself, you hardly recognized it. What do
you think the Great Unwashed heard?
>
> If we can encourage fine musicians and technicians to do and hear
> enjoyable things, re-examining their own craft in a new and exciting
> way, isn't that an important mission for anybody interested in
> non-mainstream tuning schemes? Ear-stretching beauty?

Yes, of course, but once again, as you said, you could hardly tell it
was our own temperament. How much ear stretching actually happened?
Probably not bluhdy much. How much beter to give them some Harry
Parch, or something REALLY ear stretching?
>
> As for "constructed so as to be largely overtone deficient," I thought
> the mission of modern piano builders was to create instruments that
> people enjoy playing and listening to, and buy many of...and they'll
> use whatever tonal design helps them to achieve that goal in a
> cost-effective and perhaps also artistic way. Maybe I'm mistaken in
> that naive impression.

Oi oi oi, Brad, you ARE wet behind the ears! The piano business has
been a cut-throat competition, a snake-oil peddling circus since the
Steinways starting bribing judges in the 1860's. The "Music Business"
quote often attributed to Hunter S. Thompson (which can be read by
looking at my Yahoo profile) pretty much says it all. Haven't you read
The Steinway Saga? That'll cure anyone of such romantic nonsense! The
job of modern piano manufacturers is to SURVIVE, and that means two
things: cutting costs and fighting tooth and nail for the rapidly
diminishing number of potential customers, something which they all do
more or less by spinning fanciful tales about how their instruments
are oh so much better than anyone else's, while all the time real
quality continues to descend ever more. A friend of mine in the the
Netherlands, a violinist and a piano tuner who is heavily involved in
the movement to make musicians aware of the danger to their hearing
presents by exposure to modern acoustic instruments, recently called
all the manufactures who would talk to him, and the honest ones
actually admitted that all they are trying to do is make their
instruments LOUD, 'cause that's what sells now. "Artistic" is way way
WAY down on the list, believe me...just take a look at the Steinway
website and the horribly kistchy krap they pump out in their 2art
case2 line. THAT'S what builders have to do to stay alive today.

But in any case, I was talking about the bigger trend in piano design,
which has been to dull it down since about 1830, until VERY recently
when the push to get ever more volume (or the illusion thereof) has
caused them to start introducing a harshness that never would have
been accepted even 60 years ago. They originally dulled it down to
hide the inharmonicty, and now that they are brightening it back up,
there is no miracle cure for the physics of stretched strings made of
semirigid materials.

Anyway, I'll pass your posting around to various friends in the piano
business. You can rest assured you will have brightened their day with
a little comic relief. Oh, and don't be surprised if you are soon
contacted about a certain bridge that's for sale at a very attractive
price...

;-)

Ciao,

P

As promised, here's the analysis of the harpsichord sound samples:

www.polettipiano.com/Temporary/clav002FFT.tiff

and

www.polettipiano.com/Temporary/clav002spec.tiff

Simply change the number after "clave" for each sample: 007, 011, 015,
019, 024,028, 032, and 036.

The FFT graphs indicate the average of a few seconds of the sound just
after the intensly-bright transients of the pluck have disappeared,
say about 1/3 to 1/2 second after the pluck. I watched the change of
the spectrum as the sound decayed and in every cases tried to found a
sample width and position that accurately represented the sound over
the region that we use for tuning, that is, after the initially
explosion of pluck has calmed down and before the tone has faded to a
near whisper. Actually, it is surprising as to how little the spectral
content changes as it decays. Sometimes partials come and go, as the
spectrograms indicate, and generally the higher partials fade away
leaving only the lower ones, but the FFT images pretty much indicate
what is there to hear over the most useful part of the sound.

In each case, I have indicated the position of the fundamental
frequency with the red cursor, as in the lower notes it is extremely
weak, essentially absent, because the soundboard does not have
sufficient area to radiate such low frequencies. So when looking at
the spectrograms, remember that in these instances the lowest trace is
NOT the fundamental.

I have also adjusted the screen view so that the top of the highest
peak for the strongest harmonic is just at the top of the screen. In
all cases, the space between grid lines is 5 dB, so you can easily
judge the relative strengths of harmonics among the different notes.

Right, now a sample by sample comparison of Brad's evaluation and the
objective reality.

- Clav002

Brad's analysis:

"2, normal" - it's fairly easy to pick out the 7th here and
it's well-sustained, while the 3rd and 6th harmonics (both playing the
5th degree above the root) are plenty stronger than the 7th is. The
sample becomes useless at the 10-second mark where it pulses with
unnatural noise. The 5th overtone (major 17th) makes its most
noticeable appearance about eight seconds after the pluck.

The reality:

Brad's evaluation of the relative strength of the 7th harmonic is
correct; at - 8 dB from the strongest harmonic (4), it is quite
present, and the fact that its immediate lower and upper neighbors are
slightly weaker makes it easy to hear. His comments about 3 and 6 are
only half right: 3 is stronger and 6 is slightly weaker.

- Clav007

Brad's analysis:

"1, weak" - the 3rd harmonic is so strong in there that it's
hard to hear much else, until the 5th one peeks around it after about
six or seven seconds. I can't hear any useful 7th harmonic in this tone.

The reality:

Brad fails miserably on this one. 7 is actually stronger here than in
any other sample, only about 1 dB lower than 3, the strongest. But for
all intents and purposes, there is little difference between 3, 4, 5,
6, and 7, the weakest among them, 5, being only -5 dB. 3 IS the lowest
strong harmonic. Perhaps Brad's hearing is being mislead by "lowest
component factor", meaning that the bottom of the stack is always the
easiest to hear without making any effort to focus in on specific
highest elements. 3 also maintains its level a tiny bit longer, though
6 is the clear winner in this aspect.

In any case, the fact that Brad can't hear 7 when it is so blatantly
obvious is a troubling indication of the reliability of his hearing in
this respect.

- Clav011

Brad's analysis:

"1.5, on the weak side of normal" - this tone has so much
inharmonic beating in it by itself, a conflicting resonance at the
soundboard, that it's hard to pick out much...especially among the odd
harmonics. The 2nd and 4th harmonics (octaves) are so strong and
noticeable, past the inharmonicity, that they pretty much win. I can
hear only a ghost of a 7th in there, and scarcely enough to be useful.

The Reality:

Another case where Brad's perception errs dramatically.

Strange. 7 is notably less present is this sample than in the
previous, though Brad rates it as MORE present. It is relatively weak
compared to 2, 3, 4, 5, and 6.

The most troubling aspects of Brad's analysis of this sample is his
perception of strong inharmonicty; I haven't got the foggiest notion
what he is on about here. I've done a careful harmonic by harmonic
analysis of this sample, both using very fast FFT at very high
resolution (running at 96 kHz, 3200 spectral lines, with a frequency
resolution of 0,336 Hz - not the same settings used to make the other
graphs) and aurally listening for beats with individual harmonic
frequencies, and there is absolutely no inharmonicity present in this
sound:

www.polettipiano.com/Temporary/011harmonics.mp3

(note: if you listen with small computer speakers, you probably won't
hear the fundamental and the 2nd partial in the first two plucks)

Each partial is spot-on its proper ratio to the fundamental. This
spetrogram shows a repeated loop of the Clav011 sample with momentary
overdubs of the harmonic series (bright patches) the same as the audio
file except going down through the overtone series:

www.polettipiano.com/Temporary/011inharm?.tiff

As anyone can see, there is no disagreement.

Brad: "The 2nd and 4th harmonics (octaves) are so strong and
noticeable, past the inharmonicity, that they pretty much win."

3 is actually slightly stronger than 2 and 4, with 5 being between 3
and 2, but for all intents and purposes, there is no difference
between them. Perhaps whatever is causing Brad's misconception of
inharmonicity is also blocking his ability to hear 3 and 5.

- Clav015

Brad's analysis

"2, normal" - a decent amount of 7 in there, and a good
balance all around. The strongest harmonic in here, over most of the
sustain, is 2. I can find 2, 3, 4, 5, 6, 7, and 8 in here without any
problem. What's with the chirping that starts after about six seconds?

The reality:

Very strange. This is the first sample where 7 is truly so much weaker
than the others (-29 dB) that it is essentially absent, and yet Brad
imagines he hears it. The fact that it sits in a hole, being much
lower than the harmonics just below it and just above it, means that
masking pretty much makes it impossible to hear. There is also is a
strong imbalance among the first harmonics, with 3 being noticeably
detenuated compared 1, 2, 4, and 5, despite Brad's assertion that
there is a "good balance all around."

Personal comment: For the life of me, try as I might, I can't hear 7.
If I pump up the 1 kHz region on a graphic equalizer by +24 dB (7
being about 900 Hz), I can just barely make out 7 for a fleeting
instant abut 6 seconds in. But a 24 dB boost is a LOT.

- Clav019

Brad's analysis:
"1, weak" - the strong 3 and 4 in there pretty much block out
my ability to hear much above them. I can catch a whiff of 5, but it
comes and goes.

The reality

Pretty much correct. 2 is actually the strongest, 5 is markedly weaker
at - 30 dB. 7 is actually stronger than both 5 and 6, which should
make it easier to hear than 5, if either can be heard at all.

- Clav024

Brad's analysis

"2, normal" - a decent bit of 7, but it takes a couple of
seconds to pick it out. 6 and 3 are pretty strong, and there's plenty
of 5. The inharmonicity down at the fundamental is distracting; so is
the birdie.

The reality

More or less correct on the 7. It is actually quite strong, and the
relative absence of lower neighbors make it easy to hear (for me at
least). The oddest and most troubling aspect of Brad's assertion is
that there is "plenty of 5", when in reality there is essentially no
5. At 30 dB down from its immediate lower neighbor, masking is going
to make it impossible to hear even though FFT shows that it is vaguely
(in the extreme) present.

I'm beginning to think that Brad has his own special definition of the
word "inharmonicty", much as he does with purity, since "inharmonicity
down at the fundamental" is a logical contradiction. Inharmonicty
means that the upper partials are no longer multiples of the
fundamental; inharmonicty "at the fundamental" cannot exist, since it
is THE point of reference by which any upper partial inharmonicty is
determined.

- Clav028

Brad's analysis: "1.5, weak to normal" - I can pick out 7, faintly. 2,
3, and
4 are strong and well balanced.

The reality

Regarding 7, it's more or less correct, but he's really off the mark
regarding other partials. 4 is in a really big hole, again essentially
not there because of masking from the much stronger 3. I would most
definitely not call it "well balanced".

Right. Enough is enough. Anybody who cares to look can carry on with
samples 032 and 036.

In any case, we've seen more than enough to be able to state that
Brad's hearing is just as often wrong as it is right, perhaps more
often wrong, and at times in some very significant ways. Especially
troubling are the cases of perceived inharmonicty when there is none
and the strong presence of harmonics which are actually absent or
extremely weak. Perhaps the false perception of inharmonicty is due to
a lack of proper use of terminology; now that Brad knows that there is
no true inharmonicty in these samples, maybe he could explain exactly
what the quality is that he perceives as being "inharmonic".

The point is that spectral perception is tricky business, and even
those who spend a lot of time trying to consciously do it can be
completely wrong in their judgments. I've been honing my skills with a
combined approach of listening and looking with technology for almost
30 years now, so I would like to feel that generally I can make a
pretty good guess. But I also have enough experience to know that "a
pretty good guess" is about as close as one can get. Brad, on the
other hand, seems so cocksure of himself, when actually much of his
judgment is based upon imagination or some kind of internal synthesis,
as process which has been here demonstrated to be completely
untrustworthy. It would behoove him, as it would us all, to realize
that such things can indeed influence our judgments.

Ciao,

P

Hi ... I rather suspect that either Brad, or Paul, or both, may be
wasting their time on this 'hearing test'. Specifically, the sound
that one hears may be quite different from the sound that the other
analyzes.

I do not claim to be able to judge the specific harmonic content of a
harpsichord note just by listening to it. Far from it. But I *can*
judge whether what I am hearing is a good recording or a lousy
compressed mp3.

And when I go to the 'freesound' website
http://freesound.iua.upf.edu/packsViewSingle.php?id=1491

and click on the little arrow icon above each sample, what I hear is a
lousy compressed mp3. For the first half a second of each note there
is a sort of confused burbling sound like you get on some Amazon
webpages when you want to listen to 'samples' and all they give you is
a horrid noise which sounds like no instrument on Earth. Then perhaps
a few seconds of reasonably harpsichord-like sound. Then it stops and
there is a few extra seconds of what sounds like a distant washing
machine churning away. (Where, presumably, there should be silence.)
All symptoms of a systematically abused audio file...

Now I have managed to download the .aiff files and since they are a
few meg each I assume they are good quality recordings. They are
presumably what Paul used for his analyses. (However I haven't been
able to open them ... eg they cause realplayer to abort.)

So if Brad was listening to the lousy samples on the webpage and Paul
was dealing with the full quality audio files, no wonder they come to
different conclusions.
~~~T~~~

--- In tuning@yahoogroups.com, "Paul Poletti" <paul@...> wrote:
>
> - Clav011
>
> Brad's analysis:
>
> "1.5, on the weak side of normal" - this tone has so much
> inharmonic beating in it by itself, a conflicting resonance at the
> soundboard, that it's hard to pick out much...especially among the odd
> harmonics. The 2nd and 4th harmonics (octaves) are so strong and
> noticeable, past the inharmonicity, that they pretty much win. I can
> hear only a ghost of a 7th in there, and scarcely enough to be useful.
>
> The Reality:
>
> Another case where Brad's perception errs dramatically.
>
> Strange. 7 is notably less present is this sample than in the
> previous, though Brad rates it as MORE present. It is relatively weak
> compared to 2, 3, 4, 5, and 6.
>
> The most troubling aspects of Brad's analysis of this sample is his
> perception of strong inharmonicty; I haven't got the foggiest notion
> what he is on about here. I've done a careful harmonic by harmonic
> analysis of this sample, both using very fast FFT at very high
> resolution (running at 96 kHz, 3200 spectral lines, with a frequency
> resolution of 0,336 Hz - not the same settings used to make the other
> graphs) and aurally listening for beats with individual harmonic
> frequencies, and there is absolutely no inharmonicity present in this
> sound:
>
> www.polettipiano.com/Temporary/011harmonics.mp3
>
> (note: if you listen with small computer speakers, you probably won't
> hear the fundamental and the 2nd partial in the first two plucks)
>
> Each partial is spot-on its proper ratio to the fundamental. This
> spetrogram shows a repeated loop of the Clav011 sample with momentary
> overdubs of the harmonic series (bright patches) the same as the audio
> file except going down through the overtone series:
>
> www.polettipiano.com/Temporary/011inharm?.tiff
>
> As anyone can see, there is no disagreement.
>
> Brad: "The 2nd and 4th harmonics (octaves) are so strong and
> noticeable, past the inharmonicity, that they pretty much win."
>
> 3 is actually slightly stronger than 2 and 4, with 5 being between 3
> and 2, but for all intents and purposes, there is no difference
> between them. Perhaps whatever is causing Brad's misconception of
> inharmonicity is also blocking his ability to hear 3 and 5.
>

As I said before, and I thought I said perfectly clearly: I listened
to the .aif files after first downloading the whole big zip and then
unzipping them. Clicking on them individually at the web page didn't
work from here.

As I also said clearly, going through each sample one by one, I heard
that same bizarre churning sound near the end of most of them
(although I didn't mention it every single time, because the
phenomenon got predictable and too boring to keep mentioning).

I'm still waiting for Paul to finish up his self-appointed task on
this little venture, which was apparently to provide
numerical/graphical proof (or at least the rhetorical batch of red
herrings and cherry-picked information, as pseudo-proof) that a
certain harpsichordist is unable to hear sufficiently well. Wasn't
that the point? To press the superiority of waveform analysis over
human hearing, or at least over one specific human's hearing, once and
for all?

Brad Lehman

--- In tuning@yahoogroups.com, "Tom Dent" <stringph@...> wrote:
>
>
> Hi ... I rather suspect that either Brad, or Paul, or both, may be
> wasting their time on this 'hearing test'. Specifically, the sound
> that one hears may be quite different from the sound that the other
> analyzes.
>
> I do not claim to be able to judge the specific harmonic content of a
> harpsichord note just by listening to it. Far from it. But I *can*
> judge whether what I am hearing is a good recording or a lousy
> compressed mp3.
>
> And when I go to the 'freesound' website
> http://freesound.iua.upf.edu/packsViewSingle.php?id=1491
>
> and click on the little arrow icon above each sample, what I hear is a
> lousy compressed mp3. For the first half a second of each note there
> is a sort of confused burbling sound like you get on some Amazon
> webpages when you want to listen to 'samples' and all they give you is
> a horrid noise which sounds like no instrument on Earth. Then perhaps
> a few seconds of reasonably harpsichord-like sound. Then it stops and
> there is a few extra seconds of what sounds like a distant washing
> machine churning away. (Where, presumably, there should be silence.)
> All symptoms of a systematically abused audio file...
>
> Now I have managed to download the .aiff files and since they are a
> few meg each I assume they are good quality recordings. They are
> presumably what Paul used for his analyses. (However I haven't been
> able to open them ... eg they cause realplayer to abort.)
>
> So if Brad was listening to the lousy samples on the webpage and Paul
> was dealing with the full quality audio files, no wonder they come to
> different conclusions.
> ~~~T~~~
>
>
> --- In tuning@yahoogroups.com, "Paul Poletti" <paul@> wrote:
> >
> > - Clav011
> >
> > Brad's analysis:
> >
> > "1.5, on the weak side of normal" - this tone has so much
> > inharmonic beating in it by itself, a conflicting resonance at the
> > soundboard, that it's hard to pick out much...especially among the odd
> > harmonics. The 2nd and 4th harmonics (octaves) are so strong and
> > noticeable, past the inharmonicity, that they pretty much win. I can
> > hear only a ghost of a 7th in there, and scarcely enough to be useful.
> >
> > The Reality:
> >
> > Another case where Brad's perception errs dramatically.
> >
> > Strange. 7 is notably less present is this sample than in the
> > previous, though Brad rates it as MORE present. It is relatively weak
> > compared to 2, 3, 4, 5, and 6.
> >
> > The most troubling aspects of Brad's analysis of this sample is his
> > perception of strong inharmonicty; I haven't got the foggiest notion
> > what he is on about here. I've done a careful harmonic by harmonic
> > analysis of this sample, both using very fast FFT at very high
> > resolution (running at 96 kHz, 3200 spectral lines, with a frequency
> > resolution of 0,336 Hz - not the same settings used to make the other
> > graphs) and aurally listening for beats with individual harmonic
> > frequencies, and there is absolutely no inharmonicity present in this
> > sound:
> >
> > www.polettipiano.com/Temporary/011harmonics.mp3
> >
> > (note: if you listen with small computer speakers, you probably won't
> > hear the fundamental and the 2nd partial in the first two plucks)
> >
> > Each partial is spot-on its proper ratio to the fundamental. This
> > spetrogram shows a repeated loop of the Clav011 sample with momentary
> > overdubs of the harmonic series (bright patches) the same as the audio
> > file except going down through the overtone series:
> >
> > www.polettipiano.com/Temporary/011inharm?.tiff
> >
> > As anyone can see, there is no disagreement.
> >
> > Brad: "The 2nd and 4th harmonics (octaves) are so strong and
> > noticeable, past the inharmonicity, that they pretty much win."
> >
> > 3 is actually slightly stronger than 2 and 4, with 5 being between 3
> > and 2, but for all intents and purposes, there is no difference
> > between them. Perhaps whatever is causing Brad's misconception of
> > inharmonicity is also blocking his ability to hear 3 and 5.
> >
>

--- In tuning@yahoogroups.com, "Brad Lehman" <bpl@...> wrote:

>
> As I also said clearly, going through each sample one by one, I heard
> that same bizarre churning sound near the end of most of them
> (although I didn't mention it every single time, because the
> phenomenon got predictable and too boring to keep mentioning).

Yeah, I suspect it was some sort of AGC kicking in. But so what? It
only happens near the very end of the sample, after the sound has died
away to almost nothing.
>
> I'm still waiting for Paul to finish up his self-appointed task on
> this little venture, which was apparently to provide
> numerical/graphical proof (or at least the rhetorical batch of red
> herrings and cherry-picked information, as pseudo-proof) that a
> certain harpsichordist is unable to hear sufficiently well.

Did you miss this morning's post?

/tuning/topicId_75201.html#75280

> Wasn't
> that the point?

No, the point is to give you the opportunity to learn something,
exactly as I have learned lots of times over the years about my own
sensory limitations. But then, you have to be curious enough to look
and find out rather than just think what you think you "know" is all
there is.

> To press the superiority of waveform analysis over
> human hearing, or at least over one specific human's hearing, once and
> for all?

Technology does some things better than us, others not. Determining
spectral content is just one of those things it does better. Like it
or lump it.

Ciao,

P

--- In tuning@yahoogroups.com, "Tom Dent" <stringph@...> wrote:

>
> And when I go to the 'freesound' website
> http://freesound.iua.upf.edu/packsViewSingle.php?id=1491
>
> and click on the little arrow icon above each sample, what I hear is a
> lousy compressed mp3. For the first half a second of each note there
> is a sort of confused burbling sound

Funny. Must be a Windows thing. Yeah, I hear that they are MP3's, but
there isno burbling at the beginning. Problem with Windows is there
are so many sound card options you never quite know what's gonna come out.

>Then perhaps
> a few seconds of reasonably harpsichord-like sound. Then it stops and
> there is a few extra seconds of what sounds like a distant washing
> machine churning away.

As I said, it's probably some AGC thingee trying to compensate for low
level. Not important, since as you said, it's where silence ought to be.

>
> Now I have managed to download the .aiff files and since they are a
> few meg each I assume they are good quality recordings. They are
> presumably what Paul used for his analyses.

Yep.

> (However I haven't been
> able to open them ... eg they cause realplayer to abort.)

Must be a Windows thing. Try downloading iTunes.

Ciao,

P

--- In tuning@yahoogroups.com, "Paul Poletti" <paul@...> wrote:
> Did you miss this morning's post?
>
> /tuning/topicId_75201.html#75280

Yes, I missed it. I'm downloading your slides right now to have a
closer look.

One thing I notice right away, from your explanation that accompanies
them, is that in your spectrum slides you're graphing only an snapshot
of a point where YOU thought the tone was most useful; but that I was
explicitly reporting what I hear over the whole 9 seconds (usually) of
each tone, with the various partials coming or going. So, we're not
measuring the same thing here, at all! You're expecting me to have
snapshotted a freeze-frame of my hearing at not only an isolated point
within the decay (is that even possible?), but specifically at the
SAME point you picked as important?! And with your "FFT" slides,
reporting an average, which is certainly impossible for a human ear to
calculate?

"The FFT graphs indicate the average of a few seconds of the sound
just after the intensly-bright transients of the pluck have
disappeared, say about 1/3 to 1/2 second after the pluck. I watched
the change of the spectrum as the sound decayed and in every cases
tried to found a sample width and position that accurately represented
the sound over the region that we use for tuning, that is, after the
initially explosion of pluck has calmed down and before the tone has
faded to a near whisper."

Tom's right; apples and oranges. Or maybe raw apples and the average
character of a baked apple. But, I'll take a closer look at your
graphs of your selected moments.

And I've certainly learned that it's possible to be shown "wrong" in
public by a cherry-picking approach of showing one microsecond, rather
than the whole lifetime of each tone. Geez. I thought the exercise
(along with trying to pick out the strength of 7) was to have me pick
out which harmonics I'd find most prominent and useful, if given the
opportunity hands-on to tune that specific harpsichord.

As for "inharmonicity" in my remarks, when I did your test last
weekend, I was referring to the way several of the notes were creating
a beating noise all by themselves, not being played with any other
simultaneous note. The instrument is doing something inharmonious
that is causing beating. You didn't hear that in the ones I reported,
listening all the way through the 9 seconds?

Anyway, uh, thanks.

Brad Lehman

As a newbie, I shouldn't jump in, but this was too
much fun!

I registered and downloaded the files, and listened on
a cheap iBook G4 with mediocre headphones.

The weirdness at the end of the files--the washine
machine effect--is the result of the samples being
looped, ineptly. Whoever made the recording is moving
around and breathing, etc., and this sound--with
perhaps the AGC kicking in--is being looped.

These downloaded samples at least sound cleaner than
what you hear when you click--the burbling is gone.

I second the point that an averaged spectrum, or a
snapshot of any particular point in time isn't as good
as a 3-d display that shows change over time.

And now, my audio just cut out mysteriously. Time to
say goodnight.

Caleb

--- Brad Lehman <bpl@umich.edu> wrote:

> --- In tuning@yahoogroups.com, "Paul Poletti"
> <paul@...> wrote:
> > Did you miss this morning's post?
> >
> >
>
/tuning/topicId_75201.html#75280
>
> Yes, I missed it. I'm downloading your slides right
> now to have a
> closer look.
>
> One thing I notice right away, from your explanation
> that accompanies
> them, is that in your spectrum slides you're
> graphing only an snapshot
> of a point where YOU thought the tone was most
> useful; but that I was
> explicitly reporting what I hear over the whole 9
> seconds (usually) of
> each tone, with the various partials coming or
> going. So, we're not
> measuring the same thing here, at all! You're
> expecting me to have
> snapshotted a freeze-frame of my hearing at not only
> an isolated point
> within the decay (is that even possible?), but
> specifically at the
> SAME point you picked as important?! And with your
> "FFT" slides,
> reporting an average, which is certainly impossible
> for a human ear to
> calculate?
>
> "The FFT graphs indicate the average of a few
> seconds of the sound
> just after the intensly-bright transients of the
> pluck have
> disappeared, say about 1/3 to 1/2 second after the
> pluck. I watched
> the change of the spectrum as the sound decayed and
> in every cases
> tried to found a sample width and position that
> accurately represented
> the sound over the region that we use for tuning,
> that is, after the
> initially explosion of pluck has calmed down and
> before the tone has
> faded to a near whisper."
>
> Tom's right; apples and oranges. Or maybe raw
> apples and the average
> character of a baked apple. But, I'll take a closer
> look at your
> graphs of your selected moments.
>
> And I've certainly learned that it's possible to be
> shown "wrong" in
> public by a cherry-picking approach of showing one
> microsecond, rather
> than the whole lifetime of each tone. Geez. I
> thought the exercise
> (along with trying to pick out the strength of 7)
> was to have me pick
> out which harmonics I'd find most prominent and
> useful, if given the
> opportunity hands-on to tune that specific
> harpsichord.
>
> As for "inharmonicity" in my remarks, when I did
> your test last
> weekend, I was referring to the way several of the
> notes were creating
> a beating noise all by themselves, not being played
> with any other
> simultaneous note. The instrument is doing
> something inharmonious
> that is causing beating. You didn't hear that in
> the ones I reported,
> listening all the way through the 9 seconds?
>
> Anyway, uh, thanks.
>
> Brad Lehman
>
>
>
>

--- In tuning@yahoogroups.com, "Brad Lehman" <bpl@...> wrote:

>
> I'm downloading your slides right now to have a
> closer look.
>
> One thing I notice right away, from your explanation that accompanies
> them, is that in your spectrum slides you're graphing only an snapshot
> of a point where YOU thought the tone was most useful;

Um, no, not really, you're not reading carefully. I said that I used
an averaging window which best represented the general content of the
sound over the first few seconds after the pluck. More on that point
in a moment. But the thing that I notice right away is how you start
tearing apart the analysis before you've even examined the supporting
data. How very open minded and scientific (in the best sense of the
word) of you!

> but that I was
> explicitly reporting what I hear over the whole 9 seconds (usually) of
> each tone, with the various partials coming or going.

So that means if a certain partial is there for 8 seconds and then
disappears, you say you can't hear it? Or if a partial is not there at
all, you say you CAN hear it? I don't follow...

> So, we're not
> measuring the same thing here, at all!

On the contrary, we ARE measuring the same thing here: whether or not
you can hear what is actually IN the sound at least for a pretty damn
significant portion of its totality.

> You're expecting me to have
> snapshotted a freeze-frame of my hearing at not only an isolated point
> within the decay (is that even possible?),

A red herring, Brad. I'm asking you if you heard something which was
actually there for at least a good 3 or 4 seconds (or in some cases,
showing that you claim you heard something which was NOT there over
the whole time, which right there invalidates your complaint). Your
argument is rather like when the zoo keeper runs up to you holding a
big net and says, "Hey buddy! Did you just see leopard run by here?",
you answer, "You're expecting me to have snapshotted a freeze-frame of
my sight at an isolated point within the last few moments!" Actually,
considering the speed at which these samples vary their spectral
content, it more like he asked, "Hey buddy! Did you just see an
elephant go lumbering by here?"

But as to whether or not it is even within the realm of possibility,
I'd say, yeah, it's pretty easy. Actually, I do it all the time, but
I'm only a lowly tuner, not an expert in 18th century continuo
realization.

> but specifically at the
> SAME point you picked as important?!

Well, as I said in my original post:

"Actually, it is surprising as to how little the spectral
content changes as it decays."

As you say, "one thing I notice right away" is that you neglected to
include that sentence out when you quoted my post. How very convenient
for your argument! And then you accuse ME of cherry picking?! The
juries still out as to whether or not you've got ears, Brad, but
you've sure got cojones, and big ones, too, I'll grant you that.

> And with your "FFT" slides,
> reporting an average, which is certainly impossible for a human ear to
> calculate?

Um, ah well, do you know what FFT is, Brad? Do you know how it works?
I mean, by definition, periodic oscillation is something that happens
IN TIME (which is what the word "periodic" means), so ANY manner by
which it is perceived or analyzed, human or machine, requires that it
happen OVER TIME, which is what "averaging" is also all about: the
quantification of the accumulation of similar events which occur
within a certain time frame. So, yeah, the human ear does precisely
that. It does it in a different manner, that is by a sort of
bioelectric digitization, or more accurately by running it through a
vast array of narrow bandpass filters and processing the sum total of
their combined outputs into the illusion of a pitch continuum. But it
IS a process which occurs over time, and requires a certain minimum of
time, dare we say a certain number of averages, the size/number of
which depends upon the type of signal and the frequencies contained
therein. FFT does it mathematically, but the result is more or less
the same.

Me thinks you lack sufficient technical understanding to lodge a valid
complaint against the technological approaches applied here; for
anyone with even a little experience in this field, your complaint has
about as much sophistication as "metal monster with glowing green eyes
belch fire! breath smoke! me no like metal monster!! Kill it!!"

Go read about FFT, and you will find that within the sort of time
frames we are speaking about, and with signals which vary in spectral
content over time in the manner by which these do, as demonstrated by
the spectrograms, an average of several seconds is precisely THE
method by which one acquires the most accurate representation of the
true nature of the sound as we perceive it.

Your edited version of my explanation:

> "The FFT graphs indicate the average of a few seconds of the sound
> just after the intensly-bright transients of the pluck have
> disappeared, say about 1/3 to 1/2 second after the pluck. I watched
> the change of the spectrum as the sound decayed and in every cases
> tried to found a sample width and position that accurately represented
> the sound over the region that we use for tuning, that is, after the
> initially explosion of pluck has calmed down and before the tone has
> faded to a near whisper."

So, by stating that this is some arbitrary and useless selection on my
part, you seem to be saying that you use some OTHER portion of the
sound to tune than

> a few seconds ...
> just after the intensly-bright transients of the pluck have
> disappeared

This 3 to 4 second "window of opportunity" is the point in the sound I
was taught to listen to as the easiest to hear many a long year ago by
Niel Roberts at the Harpsichord Center in LA. That's also the part of
the envelope in the sound where I still find it easiest to hear after
more than 30 years of tuning harpsichords, clavichords, and historic
pianos of all shapes and sizes (and organs, too, but they have no
envelope). Judging by the frequency with which other harpsichord
tuners repeat the notes they are tuning, that also the portion of the
envelope they all listen to. That's also the part of the envelope
where every modern piano tuner I have ever talked to listens. And
finally, that's also the part of the envelope where my students find
it easiest to pick out overtones, and not because I tell them so,
'cause often I don't have to tell them anything, they just DO it - or
already know how to do it.

But then, it wouldn't surprise me if Brad does something different
from everybody else.

There are several theoretical explanations for this common practice,
which anyone can more or less see for themselves just by looking at
the spectrograms and seeing how the sound changes over time.
Alternatively, Brad, you could go take remedial Acoustics I and
Psychoacoustics classes, if they offer such at your school. Might
clear up a lot of issues for you.

>
> And I've certainly learned that it's possible to be shown "wrong" in
> public by a cherry-picking approach of showing one microsecond, rather
> than the whole lifetime of each tone.

Mmm, now we've compressed 3 seconds out of 9 into 1 microsecond out of
a whole lifetime.

> Geez.

Indeed.

>
> As for "inharmonicity" in my remarks, when I did your test last
> weekend, I was referring to the way several of the notes were creating
> a beating noise all by themselves, not being played with any other
> simultaneous note.

Ah, exactly as I posited, evidently you have some special definition
of the word "inharmonicity" with which I am not previously acquainted.
An "inharmonic" tone does not beat by itself; self-beating (might we
call them "flagellants"?) is what is often called a "false string". I
run into them all the time on fortepianos, especially the big late 6
1/2 octave ones where they are thick as thieves. But that is not
"inharmonicty", which is something entirely different, both in sound
and what is actually occurring acoustically.

> The instrument is doing something inharmonious
> that is causing beating. You didn't hear that in the ones I reported,
> listening all the way through the 9 seconds?

Nope, I'm sorry, I can't say as if I do. Just spent 5 minutes
listening over and over again to 011 and 024 as I said before, I
haven't got the foggiest notion what you're on about here. The very
end of some of the samples have some technical problems, but so what?
The majority of the sound is free of any technical glitches which
might confuse the listener. I certainly hear nothing of any sort in
sample 011 which I would characterize as

> so much
> inharmonic beating in it by itself, a conflicting resonance at the
> soundboard, that it's hard to pick out much...especially among the
> odd harmonics.

I find the tone quite stable and far from exhibiting any sort of
falseness or inharmoncity. I also have absolutely no trouble picking
out the odd harmonics, at least the ones that are present. 3 and 5 are
both very easy to hear, requiring a minimum of focused listening. Of
course, I've been tuning historic pianos for so many years now,
perhaps I've simply learned to hear beyond any such problems.

> Anyway, uh, thanks.

Sounds like you didn't get much out of the exercise, other than
reinforcing an apparently already strong sense of innate technophobia.

Oh well, Happy Trails nonetheless.

Ciao,

P

Not sure what Paul is going on about with 'Windows'... I am currently
on a Linux system but don't have privileges. (Institute computer, guy
who controls it understandably doesn't appreciate being asked to
install things unrelated with work.)
I could try the same files in Windows if I was motivated enough to
power up another machine here.

Concerning Fourier transform, it is of course impossible to determine
the spectrum / timbre of a waveform / note over an infinitesimal
length of time, either mathematically or by ear. You have to take a
somewhat longer slice, which is if not exactly equivalent, then not a
million miles away from listening over a certain period of time.
Anyway, so far as anyone can see from the spectrum traces, the
harmonic content does *not* change all that much over time. Paul could
do a half dozen analyses on every note with different time slicing and
I doubt the results would change much.

I would say in general that the result of a mathematical harmonic
analysis, and the audible quality of a note, are two things of
different types, that cannot be directly compared in the sense of
verifying one by means of the other. One can hope to 'explain' one
partially in terms of the other, by noting general subjective
correlations, but no objective definitions of audible quality can exist.

However, Brad claimed to be able to identify partials by ear within
single harpsichord notes, and even to estimate which ones were
stronger or weaker or absent. This is an question which *can* be
answered scientifically. He was, in short, claiming to be able to
determine an objective property of the sound by ear.

If someone, listening in some 'analytic' way to the upper partials of
a tone, says that the 7th harmonic is absent or very feeble, such a
statement *can* be disproved objectively. And it seems it has been, in
a few cases.

But to me all of this is just beating about the bush. You very quickly
find out whether or not a note has a 'useful' amount of 5th or 7th
partial (etc...) by trying to hear the beating in a tempered 5/4, 7/5,
etc. etc.

You might be surprised, for instance, by how beating partials stand
out to the ear despite their low volume... Simply because they are
beating.
~~~T~~~

--- In tuning@yahoogroups.com, "Paul Poletti" <paul@...> wrote:
>
> --- In tuning@yahoogroups.com, "Tom Dent" <stringph@> wrote:
>
> >
> > And when I go to the 'freesound' website
> > http://freesound.iua.upf.edu/packsViewSingle.php?id=1491
> >
> > and click on the little arrow icon above each sample, what I hear is a
> > lousy compressed mp3. For the first half a second of each note there
> > is a sort of confused burbling sound
>
> Funny. Must be a Windows thing. Yeah, I hear that they are MP3's, but
> there isno burbling at the beginning. Problem with Windows is there
> are so many sound card options you never quite know what's gonna
come out.
>
> >Then perhaps
> > a few seconds of reasonably harpsichord-like sound. Then it stops and
> > there is a few extra seconds of what sounds like a distant washing
> > machine churning away.
>
> As I said, it's probably some AGC thingee trying to compensate for low
> level. Not important, since as you said, it's where silence ought to be.
>
> >
> > Now I have managed to download the .aiff files and since they are a
> > few meg each I assume they are good quality recordings. They are
> > presumably what Paul used for his analyses.
>
> Yep.
>
> > (However I haven't been
> > able to open them ... eg they cause realplayer to abort.)
>
> Must be a Windows thing. Try downloading iTunes.
>
> Ciao,
>
> P
>

--- In tuning@yahoogroups.com, Caleb Morgan <calebmrgn@...> wrote:
>
> As a newbie, I shouldn't jump in, but this was too
> much fun!

Why not? All opinions are welcome.
>
> I registered and downloaded the files, and listened on
> a cheap iBook G4 with mediocre headphones.

The G4 iBooks have very good audio capabilities. I had a G3 until the
video card finally died (as the all do), my girlfriend still has a G4.
I find them superior to a lot of off-the-shelf PC's in terms of audio.
>
> The weirdness at the end of the files--the washine
> machine effect--is the result of the samples being
> looped, ineptly. Whoever made the recording is moving
> around and breathing, etc., and this sound--with
> perhaps the AGC kicking in--is being looped.

yeah, sound like a good explanation

>
> I second the point that an averaged spectrum, or a
> snapshot of any particular point in time isn't as good
> as a 3-d display that shows change over time.

I agree completely that a 3D spectrogram is better than 2D FFT, but I
only have 2D software. However, I included the 2D spectrograms
precisely to show that the spectral content is quite remarkably stable
in these samples, in which case an average "snapshot" of a few seconds
is quite representative.

However, in this case I can go you even one better than a 3D, and
provide an example of the best of all possible worlds: a video of a
Very FastFT of the whole sound sample, the whole 9 yards, or 9 seconds
as the case may be, with none of the dreaded averaging (at least, none
beyond that which is inherent to the FFT process):

http://www.polettipiano.com/Temporary/Clav011movie.sitx

As with the still shots, I've set the red cursor to indicate the
fundamental frequency. It's a bit coarse on the resolution and there's
no audio, but despite these limitations, it's more than sufficient to
prove the point. Notice how surprisingly rapid the spectral content
stabilizes, and also how surprisingly long the relative levels of the
harmonics remain essentially unchanged as they begin to diminish in
strength (slide downward on the screen). About 6 seconds in, partial 4
starts to stick out above the rest (this change is easily audible),
and shortly thereafter, 5, 7 and the higher partials simultaneously
drop out of the mix. But all considered, you've got a good 6 seconds
or so when everything is more or less stable, giving the ear a good
long time to pick things out, , despite all that "inharmonicity".

In other words, in this particular case, the fact that a 3D display or
any kind of video is superior is of little importance precisely
because the nature of the signal does not require the sort of
increased visual representation regarding temporal spectral shift they
offer.

In any event, as I said, even IF there had been a lot of change going
on, the fact that Brad often reported hearing things that simply
weren't in the mix at ANY moment means that he's just grasping at
straws, desperately trying to invalidate the process on imaginary
technical grounds in order to save his credibility. This whole red
herring of "snapshots" is just more of the same.

Happy Trails!

Ciao,

P

--- In tuning@yahoogroups.com, "Tom Dent" <stringph@...> wrote:

>
> But to me all of this is just beating about the bush. You very quickly
> find out whether or not a note has a 'useful' amount of 5th or 7th
> partial (etc...) by trying to hear the beating in a tempered 5/4, 7/5,
> etc. etc.
>
> You might be surprised, for instance, by how beating partials stand
> out to the ear despite their low volume... Simply because they are
> beating.
> ~~~T~~~
>
True enuf, and in fact, if you eliminate 4 and 5 and all multiples
thereof completely from a waveform and then detune a pure major third,
it still beats, though not at all in the same way as when the critical
harmonics are present. It's very strange, actually, I'm listening to
it right now. It sounds almost as though the entire sound is rising
and falling - bah! It's making me sea-sick!

Ciao,

P

That would be really extremely weird. Well, the ear / auditory nervous
system does introduce inharmonicity at some point (as do
less-than-perfect loudspeakers/earphones etc.), so it is just
conceivable that the partials are effectively being generated at a
very low level.

Or it could be something quite different, related to some sort of
phase shifting among the partials that are actually present.

Now this is a sound file I *do* want to hear for myself. Samples please?

~~~T~~~

--- In tuning@yahoogroups.com, "Paul Poletti" <paul@...> wrote:
>
> --- In tuning@yahoogroups.com, "Tom Dent" <stringph@> wrote:
>
> > (...) beating partials stand
> > out to the ear despite their low volume... Simply because they are
> > beating.
> > ~~~T~~~
> >
> True enuf, and in fact, if you eliminate 4 and 5 and all multiples
> thereof completely from a waveform and then detune a pure major third,
> it still beats, though not at all in the same way as when the critical
> harmonics are present. It's very strange, actually, I'm listening to
> it right now. It sounds almost as though the entire sound is rising
> and falling - bah! It's making me sea-sick!
>
> Ciao,
>
> P

--- In tuning@yahoogroups.com, "Tom Dent" <stringph@...> wrote:
>
>
> That would be really extremely weird. Well, the ear / auditory nervous
> system does introduce inharmonicity at some point (as do
> less-than-perfect loudspeakers/earphones etc.), so it is just
> conceivable that the partials are effectively being generated at a
> very low level.

Actually, I think it is some sort of harmonic summing, something akin
to cumulative difference tones. I often notice that if I play a set of
frequencies with 4-5-6 proportions fairly high up (say, around 500 Hz)
and rather loudly, I hear not only 5-4=1, 6-4=2, and 6-5=1, but also a
very strong 3, which means it must be coming from the real 4 and the
two fake 1's, or the real 5 and the fake 2. I would guess that with
tones which have everything except multiples of 4 and multiples of 5,
there is some cumulative difference business going on which vaguely
fills in the illusion of beating which isn't actually there.
>
> Or it could be something quite different, related to some sort of
> phase shifting among the partials that are actually present.

Could also be...
>
> Now this is a sound file I *do* want to hear for myself. Samples please?

Right. I'll try and whip it up later this week. Right now I'm madly
cleaning up my slide show for tomorrow's class on, surprise surprise,
temperaments (for general music students).

Funny how the discussions here seem to run parallel to what I'm doing
in class, though I suspect it ends here. Next week we do the acoustics
of the organ. Anybody up for a good round of debate about the
structure of mixtures? When to use mixtures and when to use mutations?
Pipe scales? Wind pressures? Why trompettes en chamade were first
mounted horizontally? Resultant stops?

Ciao,

P

I have now got round to listening to the 'high quality' recording
files... the beginning of the notes is now cleanly recorded, and I
don't hear anything that could be called 'inharmonicity'. After the
initial pluck the sound settles down perfectly quietly (in the sense
of undisturbed) in every case.

Modulo farmyard noises, of course. I swear I heard a pig grunting in
one of the samples. Authentic to a fault.

As to the audible effects of inharmonicity, were it present, I think
they are precisely when a single note does clearly change its timbre
over a time comparable or shorter than the decay. Think of the
'Brooiiinggg' of a too-short, too-fat piano string.

I'm going to stick my neck out and say that 'false beats' are also an
example of inharmonicity. Harmonic timbre means simply that the
waveform is periodic (up to some exponential decay), with the period
corresponding to the fundamental frequency. Almost by definition, it
doesn't change in quality or beat against itself. False beats mean
that the shape of the waveform changes progressively between one
period and the next - which is what happens when the different
overtones are not multiples of the fundamental.

The stiff piano string and the false beating string are two rather
different manifestations of the same property of non-periodic
waveforms. In one case it is built into the parameters of the system,
in the other it is probably some little mechanical fault/gremlin.

But these harpsichord samples are some of the most 'straight' and, to
be frank, boring plucked tones I've come across. (I don't claim to be
able to detect their harmonic content by ear, by the way.)
~~~T~~~

Tom, do you use "false beating" synonymously with "inharmonicity"?

Oz.

----- Original Message -----
From: "Tom Dent" <stringph@gmail.com>
To: <tuning@yahoogroups.com>
Sent: 03 Mart 2008 Pazartesi 22:35
Subject: [tuning] The 'inharmonicity' again

>
> I have now got round to listening to the 'high quality' recording
> files... the beginning of the notes is now cleanly recorded, and I
> don't hear anything that could be called 'inharmonicity'. After the
> initial pluck the sound settles down perfectly quietly (in the sense
> of undisturbed) in every case.
>
> Modulo farmyard noises, of course. I swear I heard a pig grunting in
> one of the samples. Authentic to a fault.
>
> As to the audible effects of inharmonicity, were it present, I think
> they are precisely when a single note does clearly change its timbre
> over a time comparable or shorter than the decay. Think of the
> 'Brooiiinggg' of a too-short, too-fat piano string.
>
> I'm going to stick my neck out and say that 'false beats' are also an
> example of inharmonicity. Harmonic timbre means simply that the
> waveform is periodic (up to some exponential decay), with the period
> corresponding to the fundamental frequency. Almost by definition, it
> doesn't change in quality or beat against itself. False beats mean
> that the shape of the waveform changes progressively between one
> period and the next - which is what happens when the different
> overtones are not multiples of the fundamental.
>
> The stiff piano string and the false beating string are two rather
> different manifestations of the same property of non-periodic
> waveforms. In one case it is built into the parameters of the system,
> in the other it is probably some little mechanical fault/gremlin.
>
> But these harpsichord samples are some of the most 'straight' and, to
> be frank, boring plucked tones I've come across. (I don't claim to be
> able to detect their harmonic content by ear, by the way.)
> ~~~T~~~
>

--- In tuning@yahoogroups.com, "Tom Dent" <stringph@...> wrote:
>
>
> I'm going to stick my neck out and say that 'false beats' are also an
> example of inharmonicity. Harmonic timbre means simply that the
> waveform is periodic (up to some exponential decay), with the period
> corresponding to the fundamental frequency. Almost by definition, it
> doesn't change in quality or beat against itself. False beats mean
> that the shape of the waveform changes progressively between one
> period and the next - which is what happens when the different
> overtones are not multiples of the fundamental.

Watch out for them guillotine blades, Thom. They drop faster than
greased lightnin' - caaaachunk!

Actually, your mixing apples and oranges, if we can still use that
construction here. Inharmonicity is a wave form which is nonperiodic
because it has a nonharmonic spectral composition. False beating is a
periodic variation in amplitude, which is another thing altogether. It
can also be a slight periodic rise and fall in pitch. It's the same
thing as a wolf tone on violins, and it is caused by too much
compliance in the soundboard at that frequency. The board does not
present enough impedance, moves too freely with the string, and begins
to sap all the energy out of the string. As it begins to move a lot,
making the sound grow louder, it also changes the effective length of
the string, acting like either a weight or a spring attached to the
end of the string, which either effectively lengths or shortens it. In
any event, this changes the frequency of vibration and causes the
overly-efficient transfer of energy to break down, which means the
soundboard calms down again, presenting sufficient impedance to the
string to allow it to grow in vibration again, at which point the
whole cycle begins anew.

A false string may come and go with the weather, because the changes
in soundboard tension caused by expansion/contraction of the wood can
change the resonances. An inharmonic string is just inharmonic,
usually due to a material which is too stiff, or a string which is too
short compared to its diameter. Granted, irregularities in the string
material can also cause inharmonicty, precisely because localized
aberrations in mass, diameter, or stiffness (and the three are also
related) can be invisible to lower modes while quite significant at
higher modes. Too much soundboard compliance can also cause real
inharmonicty AND falseness, if the overly compliant frequency is one
of the harmonics.

But as you say, I don't hear either in any of the sounds of this
instrument. One thing Brad may be suffering from is noise pollution;
the other night I suddenly thought I WAS hearing some weird sort of
upper partial beating going on. Turned out it was just the fan on my
laptop which had kicked in because the FFT software is pretty video
intensive. Pushing the headphones against my head blocked the fan
noise and viola! no more beats.

Ciao,

P

PS BTW, did you know that the first guillotine was made by one Tobias
Schmidt, a harpsichord and fortepiano maker? He painted it black and
red, just like many a Taskin or Delin, though he didn't give it any
chinoiserri. Evidently he had a lot of black and red paint lying about
the shop, beings as the market for luxurious instruments had rather
fallen off, shall we say.

I have noticed similar effects especially if the difference tones fall in our mid hearing range. When you get to things like the fibonacci series this type of phenomena can be quite great where 2nd generation difference tones can be 'reinforced' in quite a few ways. not easily predictable beforehand though. Personally this has lead me to approach the subject of consonance and dissonance along the lines more at looking at 'acoustical coincidences', than simple ratios. The more you have, the more consonant something sounds.

Paul Poletti wrote:
>
> I often notice that if I play a set of
> frequencies with 4-5-6 proportions fairly high up (say, around 500 Hz)
> and rather loudly, I hear not only 5-4=1, 6-4=2, and 6-5=1, but also a
> very strong 3, which means it must be coming from the real 4 and the
> two fake 1's, or the real 5 and the fake 2. I would guess that with
> tones which have everything except multiples of 4 and multiples of 5,
> there is some cumulative difference business going on which vaguely
> fills in the illusion of beating which isn't actually there.
>
> Recent Activity
>
> *
> 3
> New Members
> </tuning/members;_ylc=X3oDMTJkMW1lNmhlBF9TAzk3MzU5NzE0BGdycElkAzcwNjA1BGdycHNwSWQDMTcwNTg5Nzc1MwRzZWMDdnRsBHNsawN2bWJycwRzdGltZQMxMjA0NTcxNzE5>
> *
> 1
> New Files
> </tuning/files;_ylc=X3oDMTJlYm1ydnJ1BF9TAzk3MzU5NzE0BGdycElkAzcwNjA1BGdycHNwSWQDMTcwNTg5Nzc1MwRzZWMDdnRsBHNsawN2ZmlsZXMEc3RpbWUDMTIwNDU3MTcxOQ-->
>
> Visit Your Group > </tuning;_ylc=X3oDMTJjNjIzdDF0BF9TAzk3MzU5NzE0BGdycElkAzcwNjA1BGdycHNwSWQDMTcwNTg5Nzc1MwRzZWMDdnRsBHNsawN2Z2hwBHN0aW1lAzEyMDQ1NzE3MTk-> >
> New web site?
>
> Drive traffic now. > <http://us.ard.yahoo.com/SIG=13otvfadc/M=493064.12016308.12445700.8674578/D=grplch/S=1705897753:NC/Y=YAHOO/EXP=1204578920/L=/B=NDXhJkLaX9U-/J=1204571720226893/A=3848642/R=0/SIG=131eshi2t/*http://searchmarketing.yahoo.com/arp/srchv2.php?o=US2004&cmp=Yahoo&ctv=Groups3&s=Y&s2=&s3=&b=50>
>
> Get your business
>
> on Yahoo! search.
>
> Drive Traffic
>
> Sponsored Search > <http://us.ard.yahoo.com/SIG=13oeagd9r/M=493064.12016255.12445662.8674578/D=grplch/S=1705897753:NC/Y=YAHOO/EXP=1204578920/L=/B=NTXhJkLaX9U-/J=1204571720226893/A=4025338/R=0/SIG=12jnci1fd/*http://us.rd.yahoo.com/evt=44092/*http://searchmarketing.yahoo.com/srch/index.php>
>
> can help increase
>
> your site traffic.
>
> Yahoo! Groups
>
> Women of Curves > <http://us.ard.yahoo.com/SIG=13p8lu9tt/M=493064.12016299.12445691.11322765/D=grplch/S=1705897753:NC/Y=YAHOO/EXP=1204578920/L=/B=NjXhJkLaX9U-/J=1204571720226893/A=4990219/R=0/SIG=11odsb6gn/*http://new.groups.yahoo.com/Women_Of_Curves_Everywhere>
>
> Discuss food, fitness
>
> and weight los
>
> .
--
Kraig Grady
North American Embassy of Anaphoria Island <http://anaphoria.com/index.html>
The Wandering Medicine Show
KXLU <http://www.kxlu.com/main/index.asp> 88.9 FM Wed 8-9 pm Los Angeles

Astonishingly enough, there is not a term or concept I did not understand
here.

BTW, it is spelled "chinoiserie".

Oz.

----- Original Message -----
From: "Paul Poletti" <paul@polettipiano.com>
To: <tuning@yahoogroups.com>
Sent: 03 Mart 2008 Pazartesi 23:12
Subject: [tuning] Re: The 'inharmonicity' again

> --- In tuning@yahoogroups.com, "Tom Dent" <stringph@...> wrote:
> >
> >
> > I'm going to stick my neck out and say that 'false beats' are also an
> > example of inharmonicity. Harmonic timbre means simply that the
> > waveform is periodic (up to some exponential decay), with the period
> > corresponding to the fundamental frequency. Almost by definition, it
> > doesn't change in quality or beat against itself. False beats mean
> > that the shape of the waveform changes progressively between one
> > period and the next - which is what happens when the different
> > overtones are not multiples of the fundamental.
>
> Watch out for them guillotine blades, Thom. They drop faster than
> greased lightnin' - caaaachunk!
>
> Actually, your mixing apples and oranges, if we can still use that
> construction here. Inharmonicity is a wave form which is nonperiodic
> because it has a nonharmonic spectral composition. False beating is a
> periodic variation in amplitude, which is another thing altogether. It
> can also be a slight periodic rise and fall in pitch. It's the same
> thing as a wolf tone on violins, and it is caused by too much
> compliance in the soundboard at that frequency. The board does not
> present enough impedance, moves too freely with the string, and begins
> to sap all the energy out of the string. As it begins to move a lot,
> making the sound grow louder, it also changes the effective length of
> the string, acting like either a weight or a spring attached to the
> end of the string, which either effectively lengths or shortens it. In
> any event, this changes the frequency of vibration and causes the
> overly-efficient transfer of energy to break down, which means the
> soundboard calms down again, presenting sufficient impedance to the
> string to allow it to grow in vibration again, at which point the
> whole cycle begins anew.
>
> A false string may come and go with the weather, because the changes
> in soundboard tension caused by expansion/contraction of the wood can
> change the resonances. An inharmonic string is just inharmonic,
> usually due to a material which is too stiff, or a string which is too
> short compared to its diameter. Granted, irregularities in the string
> material can also cause inharmonicty, precisely because localized
> aberrations in mass, diameter, or stiffness (and the three are also
> related) can be invisible to lower modes while quite significant at
> higher modes. Too much soundboard compliance can also cause real
> inharmonicty AND falseness, if the overly compliant frequency is one
> of the harmonics.
>
> But as you say, I don't hear either in any of the sounds of this
> instrument. One thing Brad may be suffering from is noise pollution;
> the other night I suddenly thought I WAS hearing some weird sort of
> upper partial beating going on. Turned out it was just the fan on my
> laptop which had kicked in because the FFT software is pretty video
> intensive. Pushing the headphones against my head blocked the fan
> noise and viola! no more beats.
>
> Ciao,
>
> P
>
> PS BTW, did you know that the first guillotine was made by one Tobias
> Schmidt, a harpsichord and fortepiano maker? He painted it black and
> red, just like many a Taskin or Delin, though he didn't give it any
> chinoiserri. Evidently he had a lot of black and red paint lying about
> the shop, beings as the market for luxurious instruments had rather
> fallen off, shall we say.
>

--- In tuning@yahoogroups.com, "Tom Dent" <stringph@...> wrote:
>
> That would be really extremely weird.

In some registers, some of the other upper partial pairs
can beat.

> Well, the ear / auditory
> nervous system does introduce inharmonicity at some point (as
> do less-than-perfect loudspeakers/earphones etc.),

Where's that?

> so it is just
> conceivable that the partials are effectively being generated at
> a very low level.

nah.

> Or it could be something quite different, related to some sort of
> phase shifting among the partials that are actually present.
>
> Now this is a sound file I *do* want to hear for myself.
> Samples please?
>
> ~~~T~~~
>
> --- In tuning@yahoogroups.com, "Paul Poletti" <paul@> wrote:
//
> > >
> > > True enuf, and in fact, if you eliminate 4 and 5 and all
> > > multiples thereof completely from a waveform and then
> > > detune a pure major third, it still beats,

-Carl

--- In tuning@yahoogroups.com, "Paul Poletti" <paul@...> wrote:
>
> --- In tuning@yahoogroups.com, "Tom Dent" <stringph@> wrote:
> >
> > I'm going to stick my neck out and say that 'false beats' are
> > also an example of inharmonicity. // False beats mean
> > that the shape of the waveform changes progressively between one
> > period and the next - which is what happens when the different
> > overtones are not multiples of the fundamental.
>
> Watch out for them guillotine blades, Thom. They drop faster than
> greased lightnin' - caaaachunk!
>
> Actually, your mixing apples and oranges, if we can still use that
> construction here. Inharmonicity is a wave form which is
> nonperiodic because it has a nonharmonic spectral composition.
> False beating is a periodic variation in amplitude, which is
> another thing altogether.

You're right from the spectral perspective, but in a sense Tom
is right -- false beats are a failure of harmonic motion over
the intended length of the string.

-Carl

--- In tuning@yahoogroups.com, "Carl Lumma" <carl@...> wrote:

>
> You're right from the spectral perspective, but in a sense Tom
> is right -- false beats are a failure of harmonic motion over
> the intended length of the string.
>
> -Carl
>
I disagree. The word "harmonic" means one thing and one thing only, at
least in an acoustic sense: that all spectral components are related
by being multiples of some common fundamental. It says ABSOLUTELY
nothing about amplitude, neither of any individual partial nor of the
mix taken as a whole. You can have one or more partial that
periodically varies its amplitude, and this will of course
periodicially change the wave form. This will cause what most
musicians would call "falseness" or "beating", but as long as the
frequencies of all partials remain multiples of the fundamental, the
sound remains as "harmonic" as harmonic can possibly be.

Ciao,

P

This was, really, my point. A sine wave whose amplitude is modulated,
just like two sine waves beating at a mistuned unison, is no longer a
single frequency. Let's say the false beating happens at 2Hz with a
note whose fundamental is 100Hz. Then at least some partials are
'split' or at least blurred in frequency by at least +- 1Hz.
So perhaps the 'average' of the beating partials, or their split
components, obey harmonic relations, but the sound as a whole is not
harmonic.
~~~T~~~

--- In tuning@yahoogroups.com, "Paul Poletti" <paul@...> wrote:
>
> --- In tuning@yahoogroups.com, "Carl Lumma" <carl@> wrote:
>
> >
> > You're right from the spectral perspective, but in a sense Tom
> > is right -- false beats are a failure of harmonic motion over
> > the intended length of the string.
> >
> > -Carl
> >
> I disagree. The word "harmonic" means one thing and one thing only, at
> least in an acoustic sense: that all spectral components are related
> by being multiples of some common fundamental. It says ABSOLUTELY
> nothing about amplitude, neither of any individual partial nor of the
> mix taken as a whole. You can have one or more partial that
> periodically varies its amplitude, and this will of course
> periodicially change the wave form. This will cause what most
> musicians would call "falseness" or "beating", but as long as the
> frequencies of all partials remain multiples of the fundamental, the
> sound remains as "harmonic" as harmonic can possibly be.
>
> Ciao,
>
> P

More than you ever wanted to know about false beats:

http://www.amarilli.co.uk/academic/acoustics/false_beats.pdf

-Carl

here there is ambiguity again. The composer Scelsi thought of a note as taking up more space up and down. his 4 compositions on a single note treat pitch in this manner. While ambiguous, it is compositionally useful.

Tom Dent wrote:
>
>
> This was, really, my point. A sine wave whose amplitude is modulated,
> just like two sine waves beating at a mistuned unison, is no longer a
> single frequency. Let's say the false beating happens at 2Hz with a
> note whose fundamental is 100Hz. Then at least some partials are
> 'split' or at least blurred in frequency by at least +- 1Hz.
> So perhaps the 'average' of the beating partials, or their split
> components, obey harmonic relations, but the sound as a whole is not
> harmonic.
> ~~~T~~~
>
> --- In tuning@yahoogroups.com <mailto:tuning%40yahoogroups.com>, "Paul > Poletti" <paul@...> wrote:
> >
> > --- In tuning@yahoogroups.com <mailto:tuning%40yahoogroups.com>, > "Carl Lumma" <carl@> wrote:
> >
> > >
> > > You're right from the spectral perspective, but in a sense Tom
> > > is right -- false beats are a failure of harmonic motion over
> > > the intended length of the string.
> > >
> > > -Carl
> > >
> > I disagree. The word "harmonic" means one thing and one thing only, at
> > least in an acoustic sense: that all spectral components are related
> > by being multiples of some common fundamental. It says ABSOLUTELY
> > nothing about amplitude, neither of any individual partial nor of the
> > mix taken as a whole. You can have one or more partial that
> > periodically varies its amplitude, and this will of course
> > periodicially change the wave form. This will cause what most
> > musicians would call "falseness" or "beating", but as long as the
> > frequencies of all partials remain multiples of the fundamental, the
> > sound remains as "harmonic" as harmonic can possibly be.
> >
> > Ciao,
> >
> > P
>
> -- Kraig Grady
North American Embassy of Anaphoria Island <http://anaphoria.com/index.html>
The Wandering Medicine Show
KXLU <http://www.kxlu.com/main/index.asp> 88.9 FM Wed 8-9 pm Los Angeles

I did notice myself this "false beating" caused by tuning two piano strings
with different inharmonicities to the exact same fundamental. To compensate
for the beating of an audibly strong second partial of the first string to
the second partial of the latter, the unison may be sacrificed.

Here is a no-brainer for you tuning fans:

Can a musician hear "imaginary beats" from first being subjected to a note
and then another after the first decays entirely? More precisely, does a
singer sing a "pure fifth" by taking into account beats of the upper
partials without simultaneously hearing the lower of the two pitches? Or is
the process of vocally producing intervals related more to the mental
calculation (or should I say approximation) of simple integer ratios than to
those of beats?

Prof. Ayhan Zeren claims that we can hear beats without the necessity of
simultaneously sounding tones contrary to what Helmholtz states. I found
this a peculiar incidence at the congress I am partaking. One might have
thought that tuning is precise, if not altogether possible, only when two
compound tones are sounded at the same time.

Oz.

----- Original Message -----
From: "Carl Lumma" <carl@lumma.org>
To: <tuning@yahoogroups.com>
Sent: 04 Mart 2008 Sal� 22:39
Subject: [tuning] Re: The 'inharmonicity' again

> More than you ever wanted to know about false beats:
>
> http://www.amarilli.co.uk/academic/acoustics/false_beats.pdf
>
> -Carl
>

--- In tuning@yahoogroups.com, "Ozan Yarman" <ozanyarman@...> wrote:
>
> I did notice myself this "false beating" caused by tuning
> two piano strings with different inharmonicities to the exact
> same fundamental.

That's not false beating. False beating is exhibited by
single strings!

> Here is a no-brainer for you tuning fans:
>
> Can a musician hear "imaginary beats" from first being subjected
> to a note and then another after the first decays entirely?

No, but beats between two tones presented binaurally do
are calculated by the brain.

> More precisely, does a
> singer sing a "pure fifth" by taking into account beats of
> the upper partials without simultaneously hearing the lower
> of the two pitches?

I think singers can 'hear out' partials from previously
played tones and place the fundamentals of new tones at the
same spots.

> Prof. Ayhan Zeren claims that we can hear beats without the
> necessity of simultaneously sounding tones contrary to what
> Helmholtz states. I found this a peculiar incidence at the
> congress I am partaking.

I would like to know the suggested setup for this.

-Carl

--- In tuning@yahoogroups.com, "Tom Dent" <stringph@...> wrote:
>
>
> This was, really, my point. A sine wave whose amplitude is modulated,
> just like two sine waves beating at a mistuned unison, is no longer a
> single frequency. Let's say the false beating happens at 2Hz with a
> note whose fundamental is 100Hz. Then at least some partials are
> 'split' or at least blurred in frequency by at least +- 1Hz.

I don't see the logic here. If I apply a mild tremolo to a sine wave
of 100 Hz at about 2 Hz modulation, there is no frequency shift.

You're confusing it with the beating that occurs when you combine say
98 Hz and 100 Hz, which of course beats at 2 Hz, and has a perceived
pitch of 99 Hz.

> So perhaps the 'average' of the beating partials, or their split
> components, obey harmonic relations, but the sound as a whole is not
> harmonic.

If a "component" is "split" into two theoretical components supposedly
causing the beating, then the perceived frequency is their average,
which is the same as the original frequency. Thus no frequency shift.

I can tell you from tuning a lot of fortepianos which have both false
strings and inharmonicty, the one is distinctly different from the
other. With an inharmonic tone, the overall sound appears essentially
stable in amplitude, but it is very active in timbre, changing all the
time and having a distinc metallic quality. When you try to tune it to
an octave (usually the octave above, as the bass is where we find the
most inharmonicty), you find that if you get one partial beatless, say
3, then some other partial is warbling away quit annoyingly, usually 7
for God knows whatever reason. If you stop 7 from beating, then 3 is
warbling. It is impossible to find a tuning where all the harmonics
are beatless, as is possible with most of the rest of the instrument.
Occasionally, one encounters an isolated inharmonic string higher up,
which manifests itself by causing temperament problems. I notice this
because I almost always set the temperament by matching the tones of
the tenor octave to a set of twelve tones produced by the computer,
doing it by ear. Sometimes (usually only with Italian harpsichords),
the individual unison are spot on, and the most careful listening
detects no beats at any unison. Upon checking the fifths, however, a
certain note may beat incorrectly, either way too fast or two slow. So
yo have to tweak it a bit to get the fifths beating in the right
proportion. If you return to the computer and recheck the unison, it
is out of tune slightly. Of course, the fifths in the computer beat at
the correct rates, as my tone generator (Absynth) is accurate to 6
decimal places. Luckily, I have only encountered this problem in a few
Italians harpsichords and Erard pianos of the 1840's, which are famous
for having a nasty little area of inharmonicity just below a frame
break in the scale. But in any event, I always check the temperament
by ear after matching the unisons to the computer.

But anyway, here's some samples I whipped up years ago to demonstrate
inharmonicity. The first is an harmonic tone:

http://www.polettipiano.com/inharmonicity/mild.mp3

Then a slight bit of inharmonicity:

http://www.polettipiano.com/inharmonicity/mild.mp3

and 10 times as much:

http://www.polettipiano.com/inharmonicity/intense.mp3

And finally, the intense taken down 2 octaves, which I'm sure will
remind you of the tone of strings at the bottom of a modern piano:

http://www.polettipiano.com/inharmonicity/intense-2X8ve.mp3

None of these sound anything like your typical false string, which
simple sounds as though it were beating with another string.

Enjoy!

Ciao,

P

Paul Poletti wrote:

> I don't see the logic here. If I apply a mild tremolo to a sine wave
> of 100 Hz at about 2 Hz modulation, there is no frequency shift.

I think the point is that any waveform that changes with time can't be harmonic. So it isn't so easy to make a clear definition of what a "harmonic" sound is. If you do a Fourier transform of any finite sound, it'll never be a purely harmonic signal. And if you break a waveform into individual cycles, each one will *always* be purely harmonic! So the signature of inharmonicity is that a waveform changes over time. You may well argue that tremolo isn't inharmonicity but for that you need a stricter definition than you've supplied so far.

Graham

--- In tuning@yahoogroups.com, Graham Breed <gbreed@...> wrote:
>
> Paul Poletti wrote:
>
> > I don't see the logic here. If I apply a mild tremolo to a sine wave
> > of 100 Hz at about 2 Hz modulation, there is no frequency shift.
>
> I think the point is that any waveform that changes with
> time can't be harmonic.

Not true. Consider a harmonic sound which has different rates of decay
for the different partials. At any point in time (long enough, of
course to average several periods of the fundamental), it is harmonic,
yet the shape is changing all the time. The same holds true for the
timbre changes which are produced by playing a crescendo on a wind
instrument.

> So it isn't so easy to make a clear
> definition of what a "harmonic" sound is.

Most acousticians don't seem to have much trouble. It simply means a
sound which is made up of partials which are all harmonically related
to the fundamental. It says NOTHING about their relative strengths or
whethr or not these relatively strengths are constant.

> If you do a
> Fourier transform of any finite sound, it'll never be a
> purely harmonic signal.

Well, if all harmonics are exactly in phase with the fundamental, that
is, they all start at 0 degrees at precisely the same moment, and
after precisely one complete period of the fundamental they have all
arrived again at 0 degrees, then yes, it is purely harmonic. Such
sounds rarely exist in acoustic instruments, or course, cause phase is
rarely 0 for all harmonics simultaneously. But the argument is a red
herring. Either the term is applicable to reality, as it is with my
definiton, or it's a completely useless term, in which case we'll have
to toss out the bulk a acoustics literature.

> And if you break a waveform into
> individual cycles, each one will *always* be purely
> harmonic!

If the sound is inharmonic, say like the sound of a bell, you can
never "break it into individual cycles" because there is no such thing
as "an individual cycle", only periods of the individual components. A
cycle of what?

> So the signature of inharmonicity is that a
> waveform changes over time.

Consider the singing voice, which is a harmonic instrument in that it
produces a sound composed of series of frequencies which are all
multiples of the lowest. If I begin singing an "o" and then morph it
slowly into an "a" and then into an "e", the wave form is changing all
the time, yet the sound is at all times, by the standard acoustical
defintion, "harmonic".

A bell or a bar or a stiff string, on the other hand, is never
"harmonic", precisely because the component frequencies are NOT
multiples of the fundamental.

What's the big confusion? It's so simple...

> You may well argue that tremolo
> isn't inharmonicity but for that you need a stricter
> definition than you've supplied so far.

I think I just have: see above. Correct me if I am wrong.

Ciao,

P

--- In tuning@yahoogroups.com, Graham Breed <gbreed@...> wrote:
>
> You may well argue that tremolo
> isn't inharmonicity but for that you need a stricter
> definition than you've supplied so far.

If mine doesn't convince you, try this one:

http://www.phys.unsw.edu.au/jw/sound.spectrum.html

Ciao,

P

Paul Poletti wrote:
> --- In tuning@yahoogroups.com, Graham Breed <gbreed@...> wrote:
>> You may well argue that tremolo >> isn't inharmonicity but for that you need a stricter >> definition than you've supplied so far.
> > If mine doesn't convince you, try this one:
> > http://www.phys.unsw.edu.au/jw/sound.spectrum.html

"It follows that any combination of vibrations which have frequencies made up of the harmonic series (i.e. with f, 2f, 3f, 4f, .... nf) will repeat exactly after a time T = 1/f."

Not "repeat with differing amplitude". "Repeat exactly".

Graham

Paul Poletti wrote:
> --- In tuning@yahoogroups.com, Graham Breed <gbreed@...> wrote:
>> Paul Poletti wrote:
>>
>>> I don't see the logic here. If I apply a mild tremolo to a sine wave
>>> of 100 Hz at about 2 Hz modulation, there is no frequency shift.
>> I think the point is that any waveform that changes with >> time can't be harmonic.
> > Not true. Consider a harmonic sound which has different rates of decay
> for the different partials. At any point in time (long enough, of
> course to average several periods of the fundamental), it is harmonic,
> yet the shape is changing all the time. The same holds true for the
> timbre changes which are produced by playing a crescendo on a wind
> instrument.

At any point in time, any waveform is harmonic. How do you decide if you have a harmonic sound? How long do you need to look at it, how much can it change?

>> So it isn't so easy to make a clear >> definition of what a "harmonic" sound is.
> > Most acousticians don't seem to have much trouble. It simply means a
> sound which is made up of partials which are all harmonically related
> to the fundamental. It says NOTHING about their relative strengths or
> whethr or not these relatively strengths are constant.

Maybe acousticians don't get into arguments about whether such a phenomenon is or isn't harmonic.

>> If you do a >> Fourier transform of any finite sound, it'll never be a >> purely harmonic signal.
> > Well, if all harmonics are exactly in phase with the fundamental, that
> is, they all start at 0 degrees at precisely the same moment, and
> after precisely one complete period of the fundamental they have all
> arrived again at 0 degrees, then yes, it is purely harmonic. Such
> sounds rarely exist in acoustic instruments, or course, cause phase is
> rarely 0 for all harmonics simultaneously. But the argument is a red
> herring. Either the term is applicable to reality, as it is with my
> definiton, or it's a completely useless term, in which case we'll have
> to toss out the bulk a acoustics literature.

No, it isn't purely harmonic if it exist in a finite amount of time. There must be other frequencies allowing it to come into being from nothing and go back to nothing.

When I learnt acoustics they were clear about this, and that harmonicity is an ideal that isn't fully realized in nature. It's useful in theory, like infinitely thin strings and frictionless surfaces.

>> And if you break a waveform into >> individual cycles, each one will *always* be purely >> harmonic!
> > If the sound is inharmonic, say like the sound of a bell, you can
> never "break it into individual cycles" because there is no such thing
> as "an individual cycle", only periods of the individual components. A
> cycle of what?

You can take any segment you like and call it a cycle. If a sound is slightly inharmonic, like a piano or guitar string, it will almost repeat.

>> So the signature of inharmonicity is that a >> waveform changes over time.
> > Consider the singing voice, which is a harmonic instrument in that it
> produces a sound composed of series of frequencies which are all
> multiples of the lowest. If I begin singing an "o" and then morph it
> slowly into an "a" and then into an "e", the wave form is changing all
> the time, yet the sound is at all times, by the standard acoustical
> defintion, "harmonic".

By what definition?

> A bell or a bar or a stiff string, on the other hand, is never
> "harmonic", precisely because the component frequencies are NOT
> multiples of the fundamental.

How are you measuring the component frequencies?

> What's the big confusion? It's so simple...

If you think it's simple, you're confused.

Graham

--- In tuning@yahoogroups.com, Graham Breed <gbreed@...> wrote:
>

>
> Maybe acousticians don't get into arguments about whether
> such a phenomenon is or isn't harmonic.
>
Well, I have only the experience of reading a lot of literature and
working with my colleagues here in Barcelona, most of whom are doing
pretty high-powered synthesis and/or analysis stuff over at the MTG:

http://mtg.upf.edu/

Now I know from talking to them that they have absolutely no qualms
about using the words harmonic and inharmonic to mean exactly what I
have said. I'm sure if pressed as to how to tell if a sound is
harmonic, they would say "run an FFT and see if the componenets are
multiples of the fundamental." Furthermore, I don't believe that any
of them would apply the word "inharmonic" to a sound which is shifting
gradually from one "harmonic" spectra to another via a slow change in
relative amplitudes. Since they are all pretty high-powered math
brains, I'm sure if pressed they might agree with you, but so what? As
you say, these terms have to have some basis in reality.

But I'll ask them next week at the department meeting.

Also, notice on the Uni. new South Wales page, down near the bottom:

"Here are some general statements about spectra:

* bowed strings and winds have harmonic spectra"

Now, I don't know if you've ever actually looked at the spectra of
bowed strings and winds, but they change noticeably with changes in
dynamic: generally they are more overtone rich at high dynamics, less
so at low dynamics. I don't see any caveat anywhere on this page (nor
anywhere else in the entire website, which has some pretty damn
thorough discussions of wind instrument acoustics) that either states
or implies that bowed strings and winds have harmonic spectra "only in
the limited case when the musicians maintain precisely the same
dynamic level."

Let me put it to you one final way. I can call up the spectrum window
in Absynth, and while a tone is sounding, I can increase or decrease
the amplitude of any of the individual harmonics in real time, without
having to pause the tone and star over again. Now if I made a
recording of this process, the wave form would be changing all the
time, but personally, I don't know ANYBODY who would ever say that I
was making a sounds which were inharmonic. Quite the contrary, as a
matter of fact.

Me thinks perhaps a confusion of time domain and frequency domain may
be at the center of particular bit o' angles dancing on a pinpoint
(look, another compound noun! Well well well!).

Ciao,

P

Paul wrote...
> If a "component" is "split" into two theoretical components
> supposedly causing the beating, then the perceived frequency
> is their average, which is the same as the original frequency.
> Thus no frequency shift.

The perception of the average has to do with the limited
resolution of the basilar membrane. The waveform still has
a Fourier decomposition into two sine waves.

> None of these sound anything like your typical false string, which
> simple sounds as though it were beating with another string.

I had always thought that false beating was caused by a kink
in the crystalline structure of a string, which created a node
where there weren't supposed to be none. The spectrum of the
resulting waveform would indeed not be a single harmonic
series (despite my agreeing with you on this earlier). But
the paper I linked to mentions something about a failure at
the bridge to stop one or more modes of the string. At least
I think that's what it's saying. Maybe Tom can decipher it.

-Carl

imagine if strings could be made in a weightless environment, of course only LA Monte could afford them!

Carl Lumma wrote:
>
> Paul wrote...
> > If a "component" is "split" into two theoretical components
> > supposedly causing the beating, then the perceived frequency
> > is their average, which is the same as the original frequency.
> > Thus no frequency shift.
>
> The perception of the average has to do with the limited
> resolution of the basilar membrane. The waveform still has
> a Fourier decomposition into two sine waves.
>
> > None of these sound anything like your typical false string, which
> > simple sounds as though it were beating with another string.
>
> I had always thought that false beating was caused by a kink
> in the crystalline structure of a string, which created a node
> where there weren't supposed to be none. The spectrum of the
> resulting waveform would indeed not be a single harmonic
> series (despite my agreeing with you on this earlier). But
> the paper I linked to mentions something about a failure at
> the bridge to stop one or more modes of the string. At least
> I think that's what it's saying. Maybe Tom can decipher it.
>
> -Carl
>
> -- Kraig Grady
North American Embassy of Anaphoria Island <http://anaphoria.com/index.html>
The Wandering Medicine Show
KXLU <http://www.kxlu.com/main/index.asp> 88.9 FM Wed 8-9 pm Los Angeles

Paul has shifted his ground slightly here.

I do completely believe that 'false beats' and 'inharmonicity' are, in
piano-tuning practice, essentially separate phenomena. But I am
looking for a meaningful acoustic definition, not one that relies on
the properties of pianos that Paul has prodded. We know what
'inharmonicity' means in piano-tuner jargon, but does it make
scientific sense.

First he said that a harmonic sound has partials at multiples of the
fundamental frequency. But if you take a 100Hz sinewave (let's say the
2nd harmonic of 50Hz) and amplitude modulate it at 2Hz, you get
partials at 98 and 102 Hz. All of these cannot be multiples of one
fundamental.

"The term `false beat'4 (`falseness') is a standard part of piano
tuning terminology, and refers to a beat pattern that is often present
in the decay curve of one or more audible partial tones5 of a single
string, indicating that the audible partial is not radiated from a
single normal mode, but from a *mode pair* close in frequency. Such a
pair would occur within the mode frequency series that is *otherwise*
approximately harmonic (except for some inharmonic dispersion due to
string stiffness)." (Capleton, emphasis added.)

Now Paul wants to say that the perceived frequency of each component,
where 'component' means something not yet clear, perhaps
'thing-sounding-to-me-as-if-it-were-a-partial', must be a multiple of
the fundamental. Maybe you can make a reasonable working definition
out of that too, if the components are well-behaved enough.

As to sounds with partials that gradually increase or decrease in
intensity, it does make sense to allow them to be harmonic, *if* the
intensity variations are such that they just smear out the spectrum of
each partial (e.g. Breit-Wigner lineshape, for an exponential decay)
rather than shifting or splitting it into separate parts. After all a
spectrum can never be measured to infinite precision.

My suspicion is that the basic property I want is the phase
relationships of the partials. If they do not shift relative to each
other over several periods of the fundamental, the sound is harmonic
and the features in the shape of the waveform do not shift forwards or
backwards relative to each other.

Clearly anything with 'false' or 'real' beats (acoustically almost the
same thing!) has appreciable relative phase shifts of partials.

By the way, for some reason my RealPlayer didn't accept Paul's 'mild'
mp3 file. The others ('intense' and 'intense-2X8ve') worked fine.
~~~T~~~

--- In tuning@yahoogroups.com, "Paul Poletti" <paul@...> wrote:
>
> If a "component" is "split" into two theoretical components supposedly
> causing the beating, then the perceived frequency is their average,
> which is the same as the original frequency. Thus no frequency shift.
>

Well, all this theory stuff is very fine and good. Trouble is, the two
effects do not sound alike. Listen to these two files:

http://www.polettipiano.com/inharmonicity/tremelo.mp3

http://www.polettipiano.com/inharmonicity/twotone.mp3

In both cases, the basic tone has a fundamental at 200 Hz, and is a
fairly rich tone but missing the 2nd harmonic completely. In
"tremolo", a 400 Hz sine wave has been added with a 98% amplitude
modulation at 2Hz. In "two tone" there are two tones added,
unmodulated sine waves at 399 and 401 Hz.

Now I don't know how they sound to you, but there are quite different
to me (again, listen with headphones and not some funky little
computer speakers). The tremolo just sounds like there is some beating
within the tone, whereas the two tone makes my seasick, or makes me
feel like I am in an elevator that descends too quickly, and has some
of the metallic twang of a truly inharmonic (with ALL harmonics
slightly shifted) tone, a quality which I don't hear at all in tremolo.

In any case, I think you have to admit the aural effect is no where
near identical.

Ciao,

P

404 not found
first one i can't get, even tried correcting spelling to tremolo

Paul Poletti wrote:
>
> Well, all this theory stuff is very fine and good. Trouble is, the two
> effects do not sound alike. Listen to these two files:
>
> http://www.polettipiano.com/inharmonicity/tremelo.mp3 > <http://www.polettipiano.com/inharmonicity/tremelo.mp3>
>
> http://www.polettipiano.com/inharmonicity/twotone.mp3 > <http://www.polettipiano.com/inharmonicity/twotone.mp3>
>
> In both cases, the basic tone has a fundamental at 200 Hz, and is a
> fairly rich tone but missing the 2nd harmonic completely. In
> "tremolo", a 400 Hz sine wave has been added with a 98% amplitude
> modulation at 2Hz. In "two tone" there are two tones added,
> unmodulated sine waves at 399 and 401 Hz.
>
> Now I don't know how they sound to you, but there are quite different
> to me (again, listen with headphones and not some funky little
> computer speakers). The tremolo just sounds like there is some beating
> within the tone, whereas the two tone makes my seasick, or makes me
> feel like I am in an elevator that descends too quickly, and has some
> of the metallic twang of a truly inharmonic (with ALL harmonics
> slightly shifted) tone, a quality which I don't hear at all in tremolo.
>
> In any case, I think you have to admit the aural effect is no where
> near identical.
>
> Ciao,
>
> P
>
> -- Kraig Grady
North American Embassy of Anaphoria Island <http://anaphoria.com/index.html>
The Wandering Medicine Show
KXLU <http://www.kxlu.com/main/index.asp> 88.9 FM Wed 8-9 pm Los Angeles

----- Original Message -----
From: "Carl Lumma" <carl@lumma.org>
To: <tuning@yahoogroups.com>
Sent: 05 Mart 2008 �ar�amba 9:10
Subject: [tuning] Re: The 'inharmonicity' again

> --- In tuning@yahoogroups.com, "Ozan Yarman" <ozanyarman@...> wrote:
> >
> > I did notice myself this "false beating" caused by tuning
> > two piano strings with different inharmonicities to the exact
> > same fundamental.
>
> That's not false beating. False beating is exhibited by
> single strings!
>

Ah. You mean the appearance of a shifted upper partial during decay? Where
can I hear an example?

> > Here is a no-brainer for you tuning fans:
> >
> > Can a musician hear "imaginary beats" from first being subjected
> > to a note and then another after the first decays entirely?
>
> No, but beats between two tones presented binaurally do
> are calculated by the brain.
>

So, even for psychoacoustic beats to occur, simultaneity of two tones is
crucial.

> > More precisely, does a
> > singer sing a "pure fifth" by taking into account beats of
> > the upper partials without simultaneously hearing the lower
> > of the two pitches?
>
> I think singers can 'hear out' partials from previously
> played tones and place the fundamentals of new tones at the
> same spots.
>

This is not the same as "hearing beats" obviously.

> > Prof. Ayhan Zeren claims that we can hear beats without the
> > necessity of simultaneously sounding tones contrary to what
> > Helmholtz states. I found this a peculiar incidence at the
> > congress I am partaking.
>
> I would like to know the suggested setup for this.
>

There is no setup. It appears to be an ungrounded assertion designed to
deliver the Arel-Ezgi-Uzdilek System from its incapacity to account for the
"deviant tones" of actual practice.

> -Carl
>
>

Oz.