Yahoo Tuning Groups Ultimate Backup tuning Spectrogram of a phrase in Bashir improvisation

I was discussing on facebook a clip from the Munir Bashir
improvisation I linked to in my last thread in which I swore I heard
the use of negri temperament, e.g. the equal division of 5/4 into
three steps (and then the next step being 4/3). Here's the clip I'm
talking about:

http://www.youtube.com/watch?v=A3G7VW1RMks&t=1m57s

Go to 1:57 if it doesn't automatically load. My reasoning is that he's
playing something like C Dv Eb E, but that the Eb is a bit flat to be
the 7/4 over F, so it's more like 7/6 than 6/5. The Dv seems to bisect
them, which would suggest 7-limit negri. So here's the results:

Spectrogram: http://www.mikebattagliamusic.com/music/bashirspecgram.png
Wave file: http://www.mikebattagliamusic.com/music/negri2.wav

I did the spectrogram analysis at the second harmonic which was the
strongest in the signal. The melody line is something like C Dv C Dv,
Dv Eb Dv Eb, Eb E Eb E - what I'm claiming is negri tempered. I used a
4096-point Kaiser window for the spectrogram, which means that we're
getting better frequency resolution at the cost of crappier time
resolution. This is why it looks like the note onsets are all smeared
on top of one another.

Results - The first note C is played at roughly 530 Hz; then it moves
to 572 Hz for Dv. Bashir then starts the Dv at a few Hz higher, 574
Hz, and then moves to Eb at 618 Hz. He then starts the Eb a few Hz
higher at 624 Hz, and then moves to 662 Hz. Although it sounds like
he's going to then go to F, he never does, because that's the key the
song is in and he's teasing you.

So let's work this out:

Freqs - Cents - Diffs From Previous Mean - Diffs From Previous Note
530 0.0 0.0 0.0

572 132.027 132.027
(573) (135.049) 135.049
574 138.070 6.043

618 265.937 127.867
620 271.531
(621) (274.301) 139.252
624 282.664 16.727

662 385.007 110.706 102.343

This turns outFreqs - Cents - Diffs From Previous Mean - Diffs From
Previous Note
530 0.0 0.0 0.0

572 132.027 132.027
(573) (135.049) 135.049
574 138.070 6.043

618 265.937 127.867
620 271.531
(621) (274.301) 139.252
624 282.664 16.727

662 385.007 110.706 102.343

Whether or not you want to call this a "negri well-temperament" I
leave up to you. It's also almost a perfect 12:13:14:15 tetrachord
(see for yourself).

-Mike

🔗Mike Battaglia <battaglia01@...>

Looks like the thing I had typed got erased by the chart.

The idea is that he's playing an approximate 7/6 for that minor third,
and then bisecting it. If this were exactly what he's doing, then
there'd really be no doubts at all to call this negri, as the
generators would be within 10 cents of one another.

Now, again, the melody is C Dv C Dv, Dv Eb Dv Eb, Eb E Eb E. The first
note of each new phrase is the last note of the last phrase. Sometimes
what he does is start the first note a little sharper than he ended it
the time before. And then sometimes, when he goes back to it, it ends
up lower again. In the case of the Eb at the end, there's a 20-cent
shift between the Eb at the end of Dv Eb Dv Eb and that at the start
Eb E Eb D. However, the second Eb at the latter phrase is 10 cents
flatter again.

If you take the mean frequency for all of these notes and figure out
what the step sizes are, it works out to be about 135-140-110. If the
Eb had stayed at 266, it would have been about 135-135-120, which is
even closer to Negri. A regularly-tempered Negri would see all the
generators at about 130 cents.

-Mike

On Fri, Mar 11, 2011 at 4:08 PM, Mike Battaglia <battaglia01@...> wrote:
> On Fri, Mar 11, 2011 at 3:56 PM, Mike Battaglia <battaglia01@gmail.com> wrote:
>>
>> This turns outFreqs - Cents - Diffs From Previous Mean - Diffs From
>> Previous Note
>
> Er, disregard the duplicate chart. Typo there.
>
> -Mike
>

🔗Carl Lumma <carl@...>

Mike Battaglia <battaglia01@...> wrote:

> My reasoning is that he's playing something like C Dv Eb E,
> but that the Eb is a bit flat to be the 7/4 over F,

It isn't.

> Spectrogram: http://www.mikebattagliamusic.com/music
> /bashirspecgram.png
> Wave file: http://www.mikebattagliamusic.com/music/negri2.wav
> I did the spectrogram analysis at the second harmonic which was
> the strongest in the signal. The melody line is something like
> C Dv C Dv, Dv Eb Dv Eb, Eb E Eb E - what I'm claiming is negri
> tempered. I used a 4096-point Kaiser window for the spectrogram,
> which means that we're getting better frequency resolution at
> the cost of crappier time resolution. This is why it looks like
> the note onsets are all smeared on top of one another.
> Results - The first note C is played at roughly 530 Hz; then it
> moves to 572 Hz for Dv.

All hell can break loose when you do this kind of thing.

Here it is in Tartini
http://lumma.org/temp/Tartini.png

And Melodyne
http://lumma.org/temp/Melodyne.png

Melodyne gives the first phrase as:

Bb+1, C-2, Eb-16 Eb-2, Db+55 Db+50 Db+57, Eb-27 Eb+8,

Db+55 Db+52 Db+50, Eb-34 Eb+6, Db+56 Db+56 Db+60, Eb-30 Eb-2...

Now, what scale is this? Is it negri or a harmonic tetrachord?
I don't think so! If you just look at it (or listen), it's
exactly what I said on facebook. Whether there's something
deeper about what's happening on those Eb's for instance, or
the way he lets the Dv droop before entering the second phrase,
is another question.

-Carl

🔗Mike Battaglia <battaglia01@...>

On Fri, Mar 11, 2011 at 5:29 PM, Carl Lumma <carl@...> wrote:
>
> Mike Battaglia <battaglia01@...> wrote:
>
> > My reasoning is that he's playing something like C Dv Eb E,
> > but that the Eb is a bit flat to be the 7/4 over F,
>
> It isn't.

It's a 7/6 over the C.

> > Spectrogram: http://www.mikebattagliamusic.com/music
> > /bashirspecgram.png
> > Wave file: http://www.mikebattagliamusic.com/music/negri2.wav
> > I did the spectrogram analysis at the second harmonic which was
> > the strongest in the signal. The melody line is something like
> > C Dv C Dv, Dv Eb Dv Eb, Eb E Eb E - what I'm claiming is negri
> > tempered. I used a 4096-point Kaiser window for the spectrogram,
> > which means that we're getting better frequency resolution at
> > the cost of crappier time resolution. This is why it looks like
> > the note onsets are all smeared on top of one another.
> > Results - The first note C is played at roughly 530 Hz; then it
> > moves to 572 Hz for Dv.
>
> All hell can break loose when you do this kind of thing.
>
> Here it is in Tartini
> http://lumma.org/temp/Tartini.png

I'm not sure how to interpret this. Is Tartini using the algorithm we
spoke about offlist a while ago? If so, I wouldn't trust that over a
spectrogram in a million years.

> And Melodyne
> http://lumma.org/temp/Melodyne.png
>
> Melodyne gives the first phrase as:
>
> Bb+1, C-2, Eb-16 Eb-2, Db+55 Db+50 Db+57, Eb-27 Eb+8,
>
> Db+55 Db+52 Db+50, Eb-34 Eb+6, Db+56 Db+56 Db+60, Eb-30 Eb-2...
>
> Now, what scale is this? Is it negri or a harmonic tetrachord?
> I don't think so! If you just look at it (or listen), it's
> exactly what I said on facebook. Whether there's something
> deeper about what's happening on those Eb's for instance, or
> the way he lets the Dv droop before entering the second phrase,
> is another question.

Even in your Melodyne plot, you can see that the Eb is about a third
of the way down below the of the Eb region. 1/3 of the way down would
be 266 cents, or approximately 7/6. The E is also slightly below
center, which agrees with the spectrogram I posted. I'm not sure why
it crapped out so badly on the C in the beginning.

-Mike

🔗Carl Lumma <carl@...>

Mike Battaglia <battaglia01@...> wrote:

> > > My reasoning is that he's playing something like C Dv Eb E,
> > > but that the Eb is a bit flat to be the 7/4 over F,
> >
> > It isn't.
>
> It's a 7/6 over the C.

No, it isn't.

> > Here it is in Tartini
> > http://lumma.org/temp/Tartini.png
>
> I'm not sure how to interpret this. Is Tartini using the
> algorithm we spoke about offlist a while ago? If so, I wouldn't
> trust that over a spectrogram in a million years.

Yes, and it's much better than the kind of analysis you did.

> > And Melodyne
> > http://lumma.org/temp/Melodyne.png
> > Melodyne gives the first phrase as:
> >
> > Bb+1, C-2, Eb-16 Eb-2, Db+55 Db+50 Db+57, Eb-27 Eb+8,
> >
> > Db+55 Db+52 Db+50, Eb-34 Eb+6, Db+56 Db+56 Db+60, Eb-30 Eb-2
> >
> > Now, what scale is this? Is it negri or a harmonic tetrachord?
> > I don't think so! If you just look at it (or listen), it's
> > exactly what I said on facebook. Whether there's something
> > deeper about what's happening on those Eb's for instance, or
> > the way he lets the Dv droop before entering the second phrase,
> > is another question.
>
> Even in your Melodyne plot, you can see that the Eb is
> about a third of the way down below the of the Eb region.

No, I can't. Did you even read what I wrote?

> 1/3 of the way down would
> be 266 cents, or approximately 7/6. The E is also slightly below
> center, which agrees with the spectrogram I posted.

The E is dead on.

> I'm not sure why
> it crapped out so badly on the C in the beginning.

How do you mean?

-Carl

🔗Mike Battaglia <battaglia01@...>

On Fri, Mar 11, 2011 at 6:55 PM, Carl Lumma <carl@...> wrote:
>
> Mike Battaglia <battaglia01@...> wrote:
>
> > > > My reasoning is that he's playing something like C Dv Eb E,
> > > > but that the Eb is a bit flat to be the 7/4 over F,
> > >
> > > It isn't.
> >
> > It's a 7/6 over the C.
>
> No, it isn't.

Uh huh.

> > > Here it is in Tartini
> > > http://lumma.org/temp/Tartini.png
> >
> > I'm not sure how to interpret this. Is Tartini using the
> > algorithm we spoke about offlist a while ago? If so, I wouldn't
> > trust that over a spectrogram in a million years.
>
> Yes, and it's much better than the kind of analysis you did.

The Tartini algorithm is a nonlinear, almost stochastic algorithm that
attempts to extract "the" VF from a monocomponent signal. There is no
possible way that it can be phase-invariant if it's picking peaks from
the square of its autocorrelation, if you think back to how the phase
response of each note altered the time-domain waveform of the signal
when we were looking into periodicity buzz. This algorithm is not
perfect and will introduce artifacts into the signal.

On the other hand, a spectrogram is a short-time Fourier transform,
which is a linear, reversible, deterministic algorithm that means you
chop the signal into little blocks and just take an FFT of it. It is a
linear operation that doesn't attempt to perform any type of feature
extraction on the signal it all, and hence no additional artifacts are
introduced into the signal. It doesn't assume anything about the
signal being monocomponent or do any "phantom fundamental" periodicity
processing. The tradeoff is that you have to figure out what the
fundamental at any moment is yourself. This is not a problem because
we're dealing with a harmonic signal.

The other tradeoff is that there is no "perfect" time/frequency plot;
you gain more time resolution at the cost of less frequency resolution
and vice versa. In this case, the time resolution is less important
than the frequency resolution, so I chose a window that's known for
being tight in the frequency domain and a larger window size. Vibrato
gets smeared out a bit such that the center frequency dominates.

In light of that, how on earth are you going to make the argument that
the Tartini algorithm is somehow more accurate, for this specific
purpose, than an actual STFT?

> > > And Melodyne
> > > http://lumma.org/temp/Melodyne.png
> > > Melodyne gives the first phrase as:
> > >
> > > Bb+1, C-2, Eb-16 Eb-2, Db+55 Db+50 Db+57, Eb-27 Eb+8,
> > >
> > > Db+55 Db+52 Db+50, Eb-34 Eb+6, Db+56 Db+56 Db+60, Eb-30 Eb-2
> > >
> > > Now, what scale is this? Is it negri or a harmonic tetrachord?
> > > I don't think so! If you just look at it (or listen), it's
> > > exactly what I said on facebook. Whether there's something
> > > deeper about what's happening on those Eb's for instance, or
> > > the way he lets the Dv droop before entering the second phrase,
> > > is another question.
> >
> > Even in your Melodyne plot, you can see that the Eb is
> > about a third of the way down below the of the Eb region.
>
> No, I can't. Did you even read what I wrote?

Yes, and the "first phrase" is completely screwed up. Bb C Eb? The
first phrase is "C Dv C Dv." Melodyne and Tartini both appear to have
completely crapped out on it. Where is Bb even coming from?

> > 1/3 of the way down would
> > be 266 cents, or approximately 7/6. The E is also slightly below
> > center, which agrees with the spectrogram I posted.
>
> The E is dead on.

Color me unconvinced. We know exactly what the limitations on
spectrogram analysis are and I chose a window and block size
appropriate to the type of analysis I was trying to perform. Melodyne,
on the other hand, uses a proprietary algorithm that as far as I know
is undisclosed. I have no way of determining what its limitations are,
what the common artifacts that it produces are, or how that's going to
influence the end result. It appears to have completely shredded the
first phrase beyond recognition, for starters.

-Mike

🔗Carl Lumma <carl@...>

Mike wrote:

> > > It's a 7/6 over the C.
> > No, it isn't.
> Uh huh.

No really, it isn't. The data say so, and so do my ears.
Plus there's absolutely no reason to think it would be,
since the F isn't around.

> The Tartini algorithm is a nonlinear, almost stochastic
> algorithm that attempts to extract "the" VF from a monocomponent
> signal. There is no possible way that it can be phase-invariant

What the heck are you talking about? I used two algorithms,
which just happened to agree. Melodyne uses the best pitch-
discrimination algorithm known to man. It's good enough to
use under the most stringent musical conditions.

Spectral methods tend to be too course in the range of this
analysis. Your 4096-sample window gives you a resolution of
what, about 10Hz (about 30 cents in this range).

But more importantly, you miss the big picture of what's
going on because it's too time consuming to cherry pick points
out of the spectrogram. Just look at my pngs, and you'll see
that there is both fine- and course-grain structure. The
course-grain structure is clearly 24-ET, and the fine-grain
structure is very complex, and not explained by anything like
regular temperament or extended just intonation.

> In light of that, how on earth are you going to make the
> argument that the Tartini algorithm is somehow more accurate,
> for this specific purpose, than an actual STFT?

It's way more accurate. That's why everybody who writes
software to do this kind of thing uses such methods.

>>>> And Melodyne
>>>> http://lumma.org/temp/Melodyne.png
>>>>
>>>> Bb+1, C-2, Eb-16 Eb-2, Db+55 Db+50 Db+57, Eb-27 Eb+8,
>>>>
>>>> Db+55 Db+52 Db+50, Eb-34 Eb+6, Db+56 Db+56 Db+60, Eb-30 Eb-2
>>>
>>> Even in your Melodyne plot, you can see that the Eb is
>>> about a third of the way down below the of the Eb region.
>>
>> No, I can't. Did you even read what I wrote?
>
> Yes, and the "first phrase" is completely screwed up. Bb C Eb?
> The first phrase is "C Dv C Dv." Melodyne and Tartini both
> appear to have completely crapped out on it. Where is Bb even
> coming from?

From the Bb found in the music. Here's the file, which may
start slightly before yours
http://lumma.org/temp/bashir1.wav

> > > 1/3 of the way down would
> > > be 266 cents, or approximately 7/6. The E is also slightly
> > > below center, which agrees with the spectrogram I posted.
> >
> > The E is dead on.
>
> Color me unconvinced. We know exactly what the limitations on
> spectrogram analysis are and I chose a window and block size
> appropriate to the type of analysis I was trying to perform.
> Melodyne, on the other hand, uses a proprietary algorithm that
> as far as I know is undisclosed. I have no way of determining
> what its limitations are,

It's accurate to the nearest cent in most cases because it
doesn't waste resolution up at 18KHz where you can't hear it.
We know it's damn accurate because it's been used to retune
vocal music in extended just intonation, and because Peter
Neubacker wouldn't put fake resolution into his software.

Besides which, I just verified a few of the intervals with
SPEAR, which uses STFT + spectral reassignment, the most
advanced spectral method known. It agrees to within the 5 Hz
resolution I could get out of it.

-Carl

🔗Mike Battaglia <battaglia01@...>

On Fri, Mar 11, 2011 at 8:45 PM, Carl Lumma <carl@...> wrote:
> Spectral methods tend to be too course in the range of this
> analysis. Your 4096-sample window gives you a resolution of
> what, about 10Hz (about 30 cents in this range).

Quick error - I FIR filtered and downsampled first to 22050 to speed
things up, so it's a resolution of about 5 Hz. However, I used the
Goertzel algorithm to focus on frequencies between 100 and 800 Hz that
are spaced 4 Hz apart, so I have something closer to 4 Hz resolution -
although this will basically interpolate a 4 Hz grid over the 5 Hz
background. Since we're really just trying to find the center of mass
for each frequency, I would argue that this means that we have 4 Hz
resolution for that purpose. I'll leave it at 5 Hz for the sake of
argument.

Secondly, I did all of this at the second harmonic of C partially for
this reason. The second harmonic of middle C is 522 Hz, and the second
harmonic of E above that is 658 Hz. The mean frequency here is 590 Hz.
595/590 is 14 cents, and 594/590 is 10 cents. That's closer to the
actual resolution that I have.

> > Yes, and the "first phrase" is completely screwed up. Bb C Eb?
> > The first phrase is "C Dv C Dv." Melodyne and Tartini both
> > appear to have completely crapped out on it. Where is Bb even
> > coming from?
>
> From the Bb found in the music. Here's the file, which may
> start slightly before yours
> http://lumma.org/temp/bashir1.wav

That certainly explains a lot. Can you please run it over the excerpt
I posted? The part you picked has him playing fast with a lot of
vibrato, and the wav file I used I picked specifically because he uses
less vibrato, holds the notes out a bit longer, and embellishes them
less.

http://www.mikebattagliamusic.com/music/negri2.wav

The aim is, again, not to prove that negri temperament is somehow
"fundamental" to maqam music or anything like that, but just to see if
maqam musicians on fretless instruments sometimes discover
temperaments on the fly like this during their improvisations. I swear
that there are times that that rast tetrachord is porcupine-ized.

TBH, the point of my original Facebook message was just that,
conceptually, my brain snapped what he was playing into something like
a "negri[9]" framework, as in the past have I snapped things to
various diatonic frameworks.

> > > > 1/3 of the way down would
> > > > be 266 cents, or approximately 7/6. The E is also slightly
> > > > below center, which agrees with the spectrogram I posted.
> > >
> > > The E is dead on.
> >
> > Color me unconvinced. We know exactly what the limitations on
> > spectrogram analysis are and I chose a window and block size
> > appropriate to the type of analysis I was trying to perform.
> > Melodyne, on the other hand, uses a proprietary algorithm that
> > as far as I know is undisclosed. I have no way of determining
> > what its limitations are,
>
> It's accurate to the nearest cent in most cases because it
> doesn't waste resolution up at 18KHz where you can't hear it.
> We know it's damn accurate because it's been used to retune
> vocal music in extended just intonation, and because Peter
> Neubacker wouldn't put fake resolution into his software.

Let's see how Melodyne and Tartini do over the specific clip I posted then.

> Besides which, I just verified a few of the intervals with
> SPEAR, which uses STFT + spectral reassignment, the most
> advanced spectral method known. It agrees to within the 5 Hz
> resolution I could get out of it.

What I did with Goertzel's algorithm is a very rough form of spectral
reassignment. If you have access to better algorithms (I'm not
familiar with SPEAR), I'd be satisfied. I'm still curious to see what
happens just over the specific clip I posted, however.

-Mike

🔗Carl Lumma <carl@...>

Mike Battaglia <battaglia01@...> wrote:

> Quick error - I FIR filtered and downsampled first to 22050
> to speed things up, so it's a resolution of about 5 Hz.

So about 15 cents.

> Secondly, I did all of this at the second harmonic of C
> partially for this reason.

Yep, it helps somewhat. Also the 2nd harmonic is usually
louder on the ud. But time domain methods integrate information
from all the partials, just like the ear.

> > From the Bb found in the music. Here's the file, which may
> > start slightly before yours
> > http://lumma.org/temp/bashir1.wav
>
> That certainly explains a lot. Can you please run it over the
> excerpt I posted?

Which one did you post? I used the 1:57 mark we were talking
about on facebook.

> http://www.mikebattagliamusic.com/music/negri2.wav
> The aim is, again, not to prove that negri temperament is
> somehow "fundamental" to maqam music or anything like that,
> but just to see if maqam musicians on fretless instruments
> sometimes discover temperaments on the fly like this during
> their improvisations.

I just don't see how that could possibly come about...

> Let's see how Melodyne and Tartini do over the specific clip
> I posted then.

Can you give the index into the original recording? Your
excerpt has poor fidelity. Also, please have a look at the
analysis I posted. If we can't agree on that, my doing
another isn't going to get us anywhere.

-Carl

🔗Mike Battaglia <battaglia01@...>

On Fri, Mar 11, 2011 at 10:02 PM, Carl Lumma <carl@...> wrote:
>
> > Secondly, I did all of this at the second harmonic of C
> > partially for this reason.
>
> Yep, it helps somewhat. Also the 2nd harmonic is usually
> louder on the ud. But time domain methods integrate information
> from all the partials, just like the ear.

I doubt that Melodyne is a time domain method. And I don't think that
VF extraction is necessary here. But OK.

> > That certainly explains a lot. Can you please run it over the
> > excerpt I posted?
>
> Which one did you post? I used the 1:57 mark we were talking
> about on facebook.

Try 2:05 to 2:10.

> > http://www.mikebattagliamusic.com/music/negri2.wav
> > The aim is, again, not to prove that negri temperament is
> > somehow "fundamental" to maqam music or anything like that,
> > but just to see if maqam musicians on fretless instruments
> > sometimes discover temperaments on the fly like this during
> > their improvisations.
>
> I just don't see how that could possibly come about...

I just think they decide to experiment with straightening things out a
bit. If I were a fretless musician I'd do that all the time.

> > Let's see how Melodyne and Tartini do over the specific clip
> > I posted then.
>
> Can you give the index into the original recording? Your
> excerpt has poor fidelity. Also, please have a look at the
> analysis I posted. If we can't agree on that, my doing
> another isn't going to get us anywhere.

I agree with the analysis you posted for the clip that you posted. The
part you posted was played a lot faster, which means it was probably a
bit sloppier, and had tons of vibrato on it. Specifically I think he
discovered that little figure there, realized that if he made all of
the notes 1 3/4 step intervals (or whatever) instead of alternating 1
1/2 and 2 steps or whatever, that it would sound interesting. I don't
see why not, this is improvisational music on a fretless instrument,
after all, and it's a pretty basic thing to think up.

-Mike

🔗Carl Lumma <carl@...>

🔗Mike Battaglia <battaglia01@...>

On Sat, Mar 12, 2011 at 3:42 AM, Carl Lumma <carl@...> wrote:
>
> Mike wrote:
>
> > I doubt that Melodyne is a time domain method. And I don't
> > think that VF extraction is necessary here. But OK.
>
> Melodyne DNA probably uses some spectral cues, but Melodyne
> vanilla that I'm using here is 100% time-domain.

I thought you were using DNA. Either way, I don't see why you are
suddenly of the mind that time-domain algorithms now suddenly trump
frequency-domain ones. You can probably convert one into the other
anyway.

> Since we're talking about what pitches are being used, some
> kind of VF extraction is required. You are using '1/2 the
> frequency of the 2nd harmonic' as a VF extractor.

And this avoids all of the problems that monocomponent time-domain
periodicity algorithms bring. But again, what's the point? If you have
SPEAR I'm happy enough with that, and if that matches up with Melodyne
and Tartini I'm happy with those. My protests were in that the output
didn't look anything like the clip I posted.

-Mike

🔗Mike Battaglia <battaglia01@...>

🔗Carl Lumma <carl@...>

Mike Battaglia <battaglia01@...> wrote:

> > Melodyne DNA probably uses some spectral cues, but Melodyne
> > vanilla that I'm using here is 100% time-domain.
>
> I thought you were using DNA.

Nope, DNA is for polyphonic music. :P

> Either way, I don't see why you are suddenly of the mind that
> time-domain algorithms now suddenly trump frequency-domain
> ones.

Suddenly? I've been telling you this all along! Didn't I
point you to the Dick Lyon talk about how spectral separation
in the cochlea has virtually nothing to do with how we hear?

> > > This is equivalent to saying that the field of
> > > attraction for 5/4 is much smaller for a melodic interval
> > > than a harmonic interval.
> >
> > It's equivalent to saying fields of attraction aren't the
> > One True Force Governing All Music.
>
> Can you please say whatever point it is you're trying to make
> in clear, concise English? The blind man describing an elephant
> approach isn't working here.

What makes you think fields of attraction are important in
maqam music?

> > Where did I "map" 700 cents to 3/2?
>
> You "mapped" 700 cents to 3/2 and 500 cents to 4/3 when you
> said that maqam music was in 24-tet and also that 3/2 and 4/3
> are involved because of tetrachordal symmetry.

Did not.

> > I assume they tune, for instance, the open strings as nearly
> > to 3/2 as they care to, and no more nearly to anything else.
> > The scales are tetrachordal, which certainly suggests a role
> > for 3:2. I don't know of any 5:4-period scales among the
> > maqamat - do you?
>
> I don't know of any 3:2 period scales either.

Tetrachordal constructions produces quasi-periodicity
at 3:2.

> I do know that there are plenty of trichords that I see thrown
> around all over the place, and a few are listed here, with
> outer interval 5/4:
> http://www.maqamworld.com/ajnas.html

Mike, these units are all repeated at the 4/3 or 3/2.

Also I noticed something else about this site. Care to guess
what I'm referring to?

-Carl

🔗Mike Battaglia <battaglia01@...>

On Sun, Mar 13, 2011 at 5:01 AM, Carl Lumma <carl@...> wrote:
>
> Mike Battaglia <battaglia01@...> wrote:
>
> > > Melodyne DNA probably uses some spectral cues, but Melodyne
> > > vanilla that I'm using here is 100% time-domain.
> >
> > I thought you were using DNA.
>
> Nope, DNA is for polyphonic music. :P

Oh ho ho ho! Well, during the clip I sent you, we have Omar Bashir
arpeggiating C C' C C' C C' while Munir plays the negri line above it.
And this is precisely the type of thing that is going to cause
artifacts in a monocomponent periodicity estimation algorithm.

> > Either way, I don't see why you are suddenly of the mind that
> > time-domain algorithms now suddenly trump frequency-domain
> > ones.
>
> Suddenly? I've been telling you this all along! Didn't I
> point you to the Dick Lyon talk about how spectral separation
> in the cochlea has virtually nothing to do with how we hear?

I never said that it does. I think that the cochlea splits the signal
into a number of different time-domain signals which are all then sent
to the brain. I think that the brain then performs some kind of
periodicity analysis on it. I think that the type of periodicity
analysis that it's doing is probably related to all of the
combination-sensitive neuron stuff that we've talked about before. I
think that you can model each combination-sensitive neuron as a filter
(probably a nonlinear one). This means we're back to the frequency
domain.

The fundamental question is, is the brain performing a Fourier
transform on the signal? Of course not.* However, that doesn't mean
that the frequency domain still isn't involved, nor does it mean that
it isn't involved in the cochlea which is also not performing a
Fourier transform on the signal. This is because the frequency domain
and the time domain are really the same thing, in a certain sense.

If you're filtering a signal in the frequency domain, that means that
you're multiplying the frequency response of the signal by the
frequency response of the filter. There are two ways to do this which
are mathematically equivalent:

1) Take the Fourier transform of the signal, take the Fourier
transform of the filter, multiply the two together, and then take the
inverse Fourier transform of the resultant spectrum, producing the
filtered signal
2) Take the incoming time-domain signal and in real time convolve it
with the impulse response of the filter in the time domain, producing
the filtered signal

Both the cochlea and the combination-sensitive neurons in the brain
are doing number 2 directly,** which is actually the same thing as
number 1 indirectly anyway.

> > > > This is equivalent to saying that the field of
> > > > attraction for 5/4 is much smaller for a melodic interval
> > > > than a harmonic interval.
> > >
> > > It's equivalent to saying fields of attraction aren't the
> > > One True Force Governing All Music.
> >
> > Can you please say whatever point it is you're trying to make
> > in clear, concise English? The blind man describing an elephant
> > approach isn't working here.
>
> What makes you think fields of attraction are important in
> maqam music?

The huge paragraph I wrote immediately following this one that details
the differences between your model and my model is what makes me think
that they're important in maqam music.

> > > Where did I "map" 700 cents to 3/2?
> >
> > You "mapped" 700 cents to 3/2 and 500 cents to 4/3 when you
> > said that maqam music was in 24-tet and also that 3/2 and 4/3
> > are involved because of tetrachordal symmetry.
>
> Did not.
//snip
> Tetrachordal constructions produces quasi-periodicity
> at 3:2.

aaaaaaaaaahhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh

> > I do know that there are plenty of trichords that I see thrown
> > around all over the place, and a few are listed here, with
> > outer interval 5/4:
> > http://www.maqamworld.com/ajnas.html
>
> Mike, these units are all repeated at the 4/3 or 3/2.

OK, so what?

> Also I noticed something else about this site. Care to guess
> what I'm referring to?

Yo no se.

-Mike

*Actually, you could probably turn each hair into a time-domain
integral transform if you wanted to, but that's beside the point.
**Convolution with a static, unchanging impulse response assumes that
the filters are linear and time-invariant. The filters in the cochlea
are obviously not perfectly linear, but that doesn't stop us from
pretending that they are when we talk about critical bands and stuff
like that. In the brain, I would be extremely, extremely, extremely
surprised if any of these combination-sensitive neurons display linear
behavior at all. The 1/N^2 filterbank transform model I keep proposing
is an ultra-simplification that assumes that they are. If that doesn't
turn out to be accurate, then there are nonlinear forms of convolution
that exist*** and these also have time and frequency domain
representations.
***http://folk.uio.no/ristoh/adapt/NL-filters.pdf - can I put a
footnote below another footnote? That seems dumb

🔗Carl Lumma <carl@...>

Mike Battaglia <battaglia01@...> wrote:

> Oh ho ho ho! Well, during the clip I sent you, we have
> Omar Bashir arpeggiating C C' C C' C C' while Munir plays
> the negri line above it.

Oh yeah?

> And this is precisely the type of thing that is going to cause
> artifacts in a monocomponent periodicity estimation algorithm.

Bzzz.

> > What makes you think fields of attraction are important in
> > maqam music?
>
> The huge paragraph I wrote immediately following this one that
> details the differences between your model and my model is what
> makes me think that they're important in maqam music.

Oh. I think we don't really have anything to discuss then.

> > > I do know that there are plenty of trichords that I see thrown
> > > around all over the place, and a few are listed here, with
> > > outer interval 5/4:
> > > http://www.maqamworld.com/ajnas.html
> >
> > Mike, these units are all repeated at the 4/3 or 3/2.
>
> OK, so what?

So they are not examples of anything to do with 5/4.

> > Also I noticed something else about this site. Care to guess
> > what I'm referring to?
>
> Yo no se.

The tetrachords are all defined in terms of quartertones.

-Carl

🔗Mike Battaglia <battaglia01@...>

On Sun, Mar 13, 2011 at 5:30 AM, Carl Lumma <carl@...> wrote:
>
> > And this is precisely the type of thing that is going to cause
> > artifacts in a monocomponent periodicity estimation algorithm.
>
> Bzzz.

Bzzz my ass. Go play more than one note in a pitch shift pedal
sometime that implements one of these algorithms and watch the magic
happen.

> > > What makes you think fields of attraction are important in
> > > maqam music?
> >
> > The huge paragraph I wrote immediately following this one that
> > details the differences between your model and my model is what
> > makes me think that they're important in maqam music.
>
> Oh. I think we don't really have anything to discuss then.

I take it you agree with my astute, accurate, insightful, strikingly
sapient analysis then.

> > > > I do know that there are plenty of trichords that I see thrown
> > > > around all over the place, and a few are listed here, with
> > > > outer interval 5/4:
> > > > http://www.maqamworld.com/ajnas.html
> > >
> > > Mike, these units are all repeated at the 4/3 or 3/2.
> >
> > OK, so what?
>
> So they are not examples of anything to do with 5/4.

Why is the period the only thing that matters today?

http://www.maqamworld.com/maqamat/ajam.html#ajam

The first three notes are delineated as the Ajam trichord, and then
the next four are the Kurd tetrachord starting 5/4 up from the root.

> > > Also I noticed something else about this site. Care to guess
> > > what I'm referring to?
> >
> > Yo no se.
>
> The tetrachords are all defined in terms of quartertones.

24-tet supports mohajira temperament, in other news.

-Mike

🔗Mike Battaglia <battaglia01@...>

On Sun, Mar 13, 2011 at 5:36 AM, Mike Battaglia <battaglia01@...> wrote:
>> The tetrachords are all defined in terms of quartertones.
>
> 24-tet supports mohajira temperament, in other news.

And, lastly, if you really want to be pedantic, they aren't defined in
terms of quartertones at all. The notation that they use distinguishes
between D# and Eb, and it distinguishes between D^ and D#v. This means
that neither 128/125 nor 121/120 vanish in the notation system they're
using, only 81/80. So now you're stuck with the 11-limit 81/80 and
64/63 vanishing 2D temperament.

Hell, let's be as pedantic as possible while we're at it, since this
is already an orgy of pedantism. Let's say that we're in the 2.3.~
subgroup, where ~ is an unpronounceable symbol representing a
mysterious characteristic maximum dissonance right between 6/5 and
5/4. 5, 7, and 11 don't exist at all here. So if you conflate ~^2 with
3/2, then (2~^2)/2 vanishes, bringing us down to rank-2. And now you
still end up with the Mohajira generator, and all of the maqam scales
will still be 7-note near-MOS's of this temperament. But, all of the
intervals are now given strange unpronounceable mappings. Is that
better?

-Mike

🔗lobawad <lobawad@...>

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

> > > Also I noticed something else about this site. Care to guess
> > > what I'm referring to?
> >
> > Yo no se.
>
> The tetrachords are all defined in terms of quartertones.
>
> -Carl
>

No, they are not. They are DESCRIBED, for the most part, in terms of quartertones.

Right there on page one of "ajnas" the first two sets presented are distinguished by tuning differences smaller than those afforded by quartertones, and the tuning of Busalik is described using the additional term of comma (about a 1/9 tone, or Turkish koma) variation.

Leaving aside the obvious fact that "the real" "description" of musical intervals belongs to oral tradition and let's face it, rote learning, (just as it does in Western music), the tetrachords and others "sets" at maqamworld are DEFINED in terms of internal step sizes. Which vary, as it says right there.

Description is not definition. Points of reference are not necessarily vectors of identity. A name is not who a person "is".

"Ceci n'est pas une pipe"

🔗lobawad <lobawad@...>

Heh- you posted this "in the meantime".

Yes, exactly. You're doing a great job in this discussion, Mike, nice writing and thinking!

--- In tuning@yahoogroups.com, Mike Battaglia <battaglia01@...> wrote:
>
> On Sun, Mar 13, 2011 at 5:36 AM, Mike Battaglia <battaglia01@...> wrote:
> >> The tetrachords are all defined in terms of quartertones.
> >
> > 24-tet supports mohajira temperament, in other news.
>
> And, lastly, if you really want to be pedantic, they aren't defined in
> terms of quartertones at all. The notation that they use distinguishes
> between D# and Eb, and it distinguishes between D^ and D#v. This means
> that neither 128/125 nor 121/120 vanish in the notation system they're
> using, only 81/80. So now you're stuck with the 11-limit 81/80 and
> 64/63 vanishing 2D temperament.
>
> Hell, let's be as pedantic as possible while we're at it, since this
> is already an orgy of pedantism. Let's say that we're in the 2.3.~
> subgroup, where ~ is an unpronounceable symbol representing a
> mysterious characteristic maximum dissonance right between 6/5 and
> 5/4. 5, 7, and 11 don't exist at all here. So if you conflate ~^2 with
> 3/2, then (2~^2)/2 vanishes, bringing us down to rank-2. And now you
> still end up with the Mohajira generator, and all of the maqam scales
> will still be 7-note near-MOS's of this temperament. But, all of the
> intervals are now given strange unpronounceable mappings. Is that
> better?
>
> -Mike
>

🔗lobawad <lobawad@...>

Distinguishing between description and definition is not "pedantic", it is fundamental to "understanding". Extrapolate a bit to see how heavy and important a distinction this really is.

🔗genewardsmith <genewardsmith@...>

🔗Carl Lumma <carl@...>

--- In tuning@yahoogroups.com, Mike Battaglia <battaglia01@...> wrote:

> Bzzz my ass.

Mike, the analysis I posted captures what is played perfectly.
The notes are individually inspectable in Melodyne and I only
had to correct one onset. I posted the wav file so if you want
to debate it you'd better put up something tangible.-

> I take it you agree with my astute, accurate, insightful,
> strikingly sapient analysis then.

No, I don't. To start with you should exhibit an example of
a 5/4 in a maqam recording. I don't doubt you can find one,
though as it happens, even melodic jumps of a major third
are quite rare in this music. You'd think if they were keen
on evoking a 'melodic form of periodicity detection' we would
hear examples of them playing things like C E C E C E
instead of C C' C C'...

> > So they are not examples of anything to do with 5/4.
>
> Why is the period the only thing that matters today?

You've given absolutely no reason why these outer intervals
should be 5/4. It is obvious that tetrachordal scales are
based on 3:2 intervals.

-Carl

🔗Carl Lumma <carl@...>

--- In tuning@yahoogroups.com, Mike Battaglia <battaglia01@...> wrote:

> "The Ajam trichord sounds very similar to the first 3 notes in a
> major scale in Western Classical Music, with the 3rd note tuned
> slightly lower. This makes it more mellow than a major scale."
> And then see Jiharkah, both of which are both the first three notes
> of the normal major scale:
> "The Jiharkah trichord sounds very similar to the first 3 notes in
> a major scale in Western Classical Music. The 3rd note is tuned
> slightly lower than the major scale, and even lower than in the
> Ajam trichord."

I did see those and it may be the case but multiple agreeing
sources have been notoriously difficult to find on how these
things should be intoned above the quartertone level. There's
a scholarship project for you. It should also be backed up
with analysis of recordings. Aagain, I don't doubt that 5 and 7
appear in this music (as I've said repeatedly, going back to my
analysis of the Yarman et al paper). I do doubt that they're
used in a consistent way, and I am positive that even if they
were, it still wouldn't mean regular temperament could explain
anything about this music.

-Carl

🔗Carl Lumma <carl@...>

🔗lobawad <lobawad@...>

🔗genewardsmith <genewardsmith@...>

🔗Mike Battaglia <battaglia01@...>

On Sun, Mar 13, 2011 at 2:31 PM, Carl Lumma <carl@...> wrote:
>
> > Bzzz my ass.
>
> Mike, the analysis I posted captures what is played perfectly.
> The notes are individually inspectable in Melodyne and I only
> had to correct one onset. I posted the wav file so if you want
> to debate it you'd better put up something tangible.-

I don't have Melodyne. If you'd like, you can always run all of these
expensive tools that you have on the clip that I actually posted
about.

> > I take it you agree with my astute, accurate, insightful,
> > strikingly sapient analysis then.
>
> No, I don't. To start with you should exhibit an example of
> a 5/4 in a maqam recording. I don't doubt you can find one,

Like, within how many cents? Does it have to be a 5/4 within 0.0001
cents to count as a 5/4? Or is something like 14 cents okay, so that
400 cents counts too? Why don't you just pick a random number for
error tolerance and we can go from there.

> though as it happens, even melodic jumps of a major third
> are quite rare in this music. You'd think if they were keen
> on evoking a 'melodic form of periodicity detection' we would
> hear examples of them playing things like C E C E C E
> instead of C C' C C'...

As I've said every single time that we've ever gotten into this
discussion: C Eb G, when played together, sounds "sad." You claim that
this is caused by low tonalness and low roughness. C Eb G, when
arpeggiated, sounds "sad." You claim that this same sadness is caused
by something completely different now. To an absurd extreme, you'd
claim that the Chris Potter solo I posted, which consists almost
entirely of him arpeggiating chords and the like, has a contrast
between "happy" and "sad" arpeggiations for some mysterious reason
that has nothing to do with tonalness that you have never explained.
What do you claim it's caused by? What evidence do you have to back up
your claim that this is a separate mechanism?

> > > So they are not examples of anything to do with 5/4.
> >
> > Why is the period the only thing that matters today?
>
> You've given absolutely no reason why these outer intervals
> should be 5/4. It is obvious that tetrachordal scales are
> based on 3:2 intervals.

Yes, here's the obvious reason: 400 cents sounds like 5/4. If you'd
like, we can go run an mTurk test where we have two notes played one
after the other and see how well HE predicts whether or not people say
they're consonant or dissonant.

-Mike

🔗Mike Battaglia <battaglia01@...>

🔗Carl Lumma <carl@...>

Mike Battaglia <battaglia01@...> wrote:

> > No, I don't. To start with you should exhibit an example of
> > a 5/4 in a maqam recording. I don't doubt you can find one,
>
> Like, within how many cents?

Some relevant question are

* Are intervals within 5:4's field of attraction treated
as consonances in the music?

* Of the intervals performed, what fraction are within
5:4's field of attraction?

* For intervals performed within 5:4's filed of attraction,
are the distributed in some sort of suggestive fashion
around 386 cents?

etc etc.

> As I've said every single time that we've ever gotten into
> this discussion: C Eb G, when played together, sounds "sad."

I fail to see the relevance of the minorness topic here.

> C Eb G, when arpeggiated, sounds "sad."

Does it? Wait, don't answer that.

-Carl

🔗Mike Battaglia <battaglia01@...>

On Sun, Mar 13, 2011 at 5:09 PM, Carl Lumma <carl@...> wrote:
>
> > > No, I don't. To start with you should exhibit an example of
> > > a 5/4 in a maqam recording. I don't doubt you can find one,
> >
> > Like, within how many cents?
>
> Some relevant question are
>
> * Are intervals within 5:4's field of attraction treated
> as consonances in the music?

Define "treated as consonances."

> * Of the intervals performed, what fraction are within
> 5:4's field of attraction?

Assuming we're going back to the usual HE field of attraction for 5/4,
then all of the 400 cent ones are.

> * For intervals performed within 5:4's filed of attraction,
> are the distributed in some sort of suggestive fashion
> around 386 cents?

They aren't for music in 12-tet either.

> > As I've said every single time that we've ever gotten into
> > this discussion: C Eb G, when played together, sounds "sad."
>
> I fail to see the relevance of the minorness topic here.

C D E, when arpeggiated, sounds "happy." Maqam musicians arpeggiate
this all the time. There's your relevance.

> > C Eb G, when arpeggiated, sounds "sad."
>
> Does it?

Yes.

> Wait, don't answer that.

This is what we call "treading around the elephant in the room."

-Mike

🔗Carl Lumma <carl@...>

--- In tuning@yahoogroups.com, Mike Battaglia <battaglia01@...> wrote:
>
> > * Are intervals within 5:4's field of attraction treated
> > as consonances in the music?
>
> Define "treated as consonances."

It's your idea; you have to convince us of this. Personally
I don't think it'll be very easy but there you go.

> > * Of the intervals performed, what fraction are within
> > 5:4's field of attraction?
>
> Assuming we're going back to the usual HE field of attraction
> for 5/4, then all of the 400 cent ones are.

Of *all* intervals performed, what fraction are.

> > * For intervals performed within 5:4's filed of attraction,
> > are the distributed in some sort of suggestive fashion
> > around 386 cents?
>
> They aren't for music in 12-tet either.

Of course they are! Well, they're feeling the pull of
81/64 too, so the distribution is probably north of 386.
But we know they hit 386 more or less exactly a fair
amount of the time.

> This is what we call "treading around the elephant in the room."

You are treading around the elephant of the very dead
simple picture based on clear facts that I have presented,
with totally baseless speculation about negri temperament.

-Carl

🔗Mike Battaglia <battaglia01@...>

🔗Carl Lumma <carl@...>

🔗Mike Battaglia <battaglia01@...>

On Sun, Mar 13, 2011 at 5:55 PM, Carl Lumma <carl@...> wrote:
>
> --- In tuning@yahoogroups.com, Mike Battaglia <battaglia01@...> wrote:
> >
> > > * Are intervals within 5:4's field of attraction treated
> > > as consonances in the music?
> >
> > Define "treated as consonances."
>
> It's your idea; you have to convince us of this. Personally
> I don't think it'll be very easy but there you go.

They sound happy when arpeggiated. Happiness has something to do with
HE. As a heuristic, it seems plausible that HE then applies to melodic
intervals. For more rigorous science, I have proposed the mTurk test.

> > > * Of the intervals performed, what fraction are within
> > > 5:4's field of attraction?
> >
> > Assuming we're going back to the usual HE field of attraction
> > for 5/4, then all of the 400 cent ones are.
>
> Of *all* intervals performed, what fraction are.

You're saying that you think they'll often play those sharp? If so,
then the semififths dominant temperament seems like a good choice
again.

> > > * For intervals performed within 5:4's filed of attraction,
> > > are the distributed in some sort of suggestive fashion
> > > around 386 cents?
> >
> > They aren't for music in 12-tet either.
>
> Of course they are! Well, they're feeling the pull of
> 81/64 too, so the distribution is probably north of 386.
> But we know they hit 386 more or less exactly a fair
> amount of the time.

Playing intervals simultaneously, for harmonic timbres where beating
is present, can compel people to microtune intervals closer to just to
avoid the beating. This will obviously not be an impetus for intervals
played in succession. But this also has nothing to do with my claim,
which is that the perception of 400 cents when played as a melodic
interval is related to the field of attraction for 5/4 through a
mixture, roughly put, of harmonic entropy and priming.

It is also why I keep saying that the "low roughness" part of your
desideratum for minorness isn't relevant for arpeggiated intervals,
which is because there's no cochlear interaction at all if they're
arpeggiated.

> > This is what we call "treading around the elephant in the room."
>
> You are treading around the elephant of the very dead
> simple picture based on clear facts that I have presented,
> with totally baseless speculation about negri temperament.

Tartini data coming up on this.

But I fail to see how negri temperament is related to the present
discussion. I also fail to see why you're blowing this issue up out of
the context that I posted it in. I said that it sounded like Munir
Bashir had evened out the steps in a tiny fragment in an improvisation
played on a fretless instrument. I informally referred to this pattern
as "Negri" on Facebook, where I got in a tangential discussion with
Paul over the rigor of using that terminology. I never claimed that
maqam music was "negri tempered" or that negri plays any part in its
structure. I claimed, initially, that as a one-off incident in an
improvisation, it sounded like a pattern of steps was occurring that I
related to the negri 9-note MOS.

In fact, I never even claimed that it actually was negri tempered,
just that it sounded like that to me, subjectively, on the recording.
If you really want me to be accurate, the whole point of that
discussion was "Hey Igs look, this sounds like 'negri hearing' instead
of 'diatonic hearing!'" And when I can figure out how to use Tartini
we'll see how good my ears are. Maybe you really are golden ears here,
and all of the infamous AP stuff I can pull in real life only works in
12-tet, kind of how like chess players apparently have trouble keeping
track of a game played mentally if the pieces are set up in an
impossible position.

This is all a side discussion. The present discussion is about whether
or not you can map intervals in maqam music to 5 and 11. You claim
that we can't since they're used as melodic intervals, and hence the
usual periodicity analysis isn't taking place, so it's inappropriate
to map 400 cents to 5/4. The undercurrent here is the same
disagreement we've had in every major discussion in the past year,
which is whether or not HE applies to melodic processing.

This is why all of the stuff is related on minorness. This is also why
my "very dead simple picture" of the linear temperaments suggested is
a viable springboard onto further research. Claiming that it's not
accurate enough, or that intervals are being mapped inconsistently, or
something like that, would be a good criticism of those temperaments.
Claiming that mapping to primes is an invalid procedure for monophonic
music is the focus of the present conversation.

-Mike

🔗lobawad <lobawad@...>

--- In tuning@yahoogroups.com, Mike Battaglia <battaglia01@...> wrote:
>
> On Sun, Mar 13, 2011 at 5:03 PM, Carl Lumma <carl@...> wrote:
> >
> > --- In tuning@yahoogroups.com, Mike Battaglia <battaglia01@> wrote:
> >
> > > I see you're starting the ridiculous strawman phase of the discussion
> > > now. Sweet.
> >
> > Cameron said he thought you were doing a great job. I was
> > wondering he also agrees with your use of negri temperament to
> > characterize this recording. -Carl
>
> Oh, so that's what you were wondering! I didn't gather that from what you said.
>
> -Mike
>

I didn't gather that, either. Looking over the thread as much as possible given the dreadful threading scheme of Yahoo groups, I see that wat you DID propose, correct me if I'm wrong, is that we could DESCRIBE a certain maqam performance in terms of "negri temperament".

I don't know what "negri" temperament is, but yes of course it is possible to describe all kinds of musical performances in terms of regular temperaments.

I assume, given your numerous workings-out of commatic action in terms of chord progressions, that what you mean is "usefully describe".

🔗Mike Battaglia <battaglia01@...>

On Sun, Mar 13, 2011 at 6:11 PM, Carl Lumma <carl@...> wrote:
>
> --- In tuning@yahoogroups.com, Mike Battaglia <battaglia01@...> wrote:
>
> > And, I should put, my offer to get an mTurk test set up still
> > stands. If HE doesn't apply to melodies, then if we do a
> > listening test with two notes with sines, played one after the
> > other, in which people go through the HE minima and maxima in
> > order, then their rating of the concordance of these adjacent
> > intervals should be no more than chance, right?
>
> I have no idea what you mean, or what hypothesis you're
> trying to test. I am all in favor of mTurk experiments.
> For now I've got the triads URLs to send, Bashir to analyze,
> family to be with, and intergalactic-size bullshit canons
> on Fukushima and maqam music to defuse. It's been an
> interesting weekend. -Carl

HE's validity, according to you, has been tested through a lot of
informal subjective listening tests over the concordance and
discordance of different harmonic intervals. That is as valid footing
as it stands on. I want to test its validity in describing melodic
intervals played one after the other. That way, when I do, then I
won't be hearing any more nonsense about how we can't map 400 cents to
5/4.

So I want to do a listening test with two notes with sines, played one
after the other, in which people go through the HE minima and maxima
in random order and rate their perceived consonance or dissonance.
Some kind of preliminary introduction phase so that people can
understand what we're talking about would probably be good. If HE
doesn't apply to melodies, these ratings should be pretty random,
right?

-Mike

🔗Mike Battaglia <battaglia01@...>

On Sun, Mar 13, 2011 at 6:23 PM, lobawad <lobawad@...> wrote:
>
> I didn't gather that, either. Looking over the thread as much as possible given the dreadful threading scheme of Yahoo groups, I see that wat you DID propose, correct me if I'm wrong, is that we could DESCRIBE a certain maqam performance in terms of "negri temperament".

This discussion stemmed out of a post I made on Facebook. I was in
Munir Bashir meditative trance land, and heard him play C Dv Eb E over
and over and over. As he kept playing it, it sounded like he started
to straighten the intervals out such that the neutral second and the
half step all became one middle of the road second, aka an equal
division of 5/4 into three parts. And then it sounded like the F,
which would follow, would be another middle of the road second of the
same size, such that 4/3 is divided into four equal parts in such a
way that the third is 5/4. This is part of the 9-note Negri MOS.

I then informally said that this clip sounded like "straight up negri
temperament," which really meant "wow! I listened to maqam music and
it not only destroyed my learned sense of diatonic hearing but seems
to have activated a new one, which after searching my memory is
apparently related to negri temperament! Yay!" Then we went in circles
a bit about whether or not you're allowed to call that negri
temperament if you want to be 100% rigorous, which I think that you
are. And now we're on about whether you can map primes to melodic
intervals at all.

Whether the clip turns out to actually be negri-tempered or not
doesn't mean anything, because it was a statement about my categorical
perception at the time, not a claim about the fundamental underlying
structure of maqam music.

-Mike

🔗Carl Lumma <carl@...>

--- In tuning@yahoogroups.com, Mike Battaglia <battaglia01@...> wrote:

> > > > * Of the intervals performed, what fraction are within
> > > > 5:4's field of attraction?
> > >
> > > Assuming we're going back to the usual HE field of attraction
> > > for 5/4, then all of the 400 cent ones are.
> >
> > Of *all* intervals performed, what fraction are.
>
> You're saying that you think they'll often play those sharp?

Of *all* intervals performed, what fraction are.

> This will obviously not be an impetus for intervals
> played in succession. But this also has nothing to do with my claim,
> which is that the perception of 400 cents when played as a melodic
> interval is related to the field of attraction for 5/4 through a
> mixture, roughly put, of harmonic entropy and priming.

Definitely not priming, for the 5th time.

> I informally referred to this pattern
> as "Negri" on Facebook,

You've very strongly asserted it here, quite recently in fact.

> This is all a side discussion. The present discussion is about
> whether or not you can map intervals in maqam music to 5 and 11.
> You claim that we can't since they're used as melodic intervals,

No, that is not what I've claimed. I think you need to
take a breather, Mike.

-Carl

🔗Mike Battaglia <battaglia01@...>

On Sun, Mar 13, 2011 at 6:34 PM, Carl Lumma <carl@...> wrote:
>
> > You're saying that you think they'll often play those sharp?
>
> Of *all* intervals performed, what fraction are.

This is silly. 435 cents is in 5/4's field of attraction, as you are
fond of pointing out. You really require me to test to confirm that
the major thirds are within this absurdly large range?

> > This will obviously not be an impetus for intervals
> > played in succession. But this also has nothing to do with my claim,
> > which is that the perception of 400 cents when played as a melodic
> > interval is related to the field of attraction for 5/4 through a
> > mixture, roughly put, of harmonic entropy and priming.
>
> Definitely not priming, for the 5th time.

See "roughly put."

> > I informally referred to this pattern
> > as "Negri" on Facebook,
>
> You've very strongly asserted it here, quite recently in fact.

I never asserted that Negri had anything fundamental to do with the
structure of maqam music. I said, and I don't know how much longer I
have to keep saying it, that as a one off thing, it sounded like Munir
had decided to straighten the steps of C Dv Eb E out. I have said very
clearly that I thought he was inspired in the middle of his
improvisation to do so, and that it has nothing to do with the
fundamental underpinnings of maqam structure. In fact, I also said
that this was just the categorical perception I had of the phrase, and
that it may well be closer to 0-150-300-400. And, if you at one point
thought that I was saying anything other than this, then let this set
the record straight, so I can stop hearing this same strawman argument
made over and over.

> > This is all a side discussion. The present discussion is about
> > whether or not you can map intervals in maqam music to 5 and 11.
> > You claim that we can't since they're used as melodic intervals,
>
> No, that is not what I've claimed. I think you need to
> take a breather, Mike.

This may be because when you make a claim, you make it by claiming
various fourth-order implications of your idea without just stating
what the idea is. I have written several very long paragraphs to try
and figure out precisely what it is that you are claiming, and every
time I do, you don't include them in your reply. I can only assume
that you aren't even reading them. What is the purpose of
communication at this point?

Here are some quotes that you've made in the present conversation:

> Sigh... I see you've decided to ignore Paul on this.
> "11-limit intervals"... "temperament"... you haven't justified
the use of either of these terms.

> What makes you think fields of attraction are important in maqam music?

> You're conflating harmonic and melodic listening. Trained
> recognition of quantitative values with naive perception
> of qualia. Just for starters.

> The scales are tetrachordal, which certainly suggests a role
> for 3:2. I don't know of any 5:4-period scales among the
> maqamat - do you?

Which, after I sent out some 5/4-period ajnas, you raised the goalpost
and said that although the period here is 400 cents, that that doesn't
count as the period being "5/4," for some mysterious reason.

> You'd think if they were keen
> on evoking a 'melodic form of periodicity detection' we would
> hear examples of them playing things like C E C E C E
> instead of C C' C C'...

etc. I spent a lot of time trying to write detailed analyses of what I
understand our two paradigms to be, what the difference between them
is, and how I think we can figure out which is more accurate with an
actual listening test. I hoped you'd respond to these, because
otherwise I might as well be talking to you in Chinese. You didn't, so
this is the best view I have of what you're saying. What is the view,
if not?

-Mike

🔗Carl Lumma <carl@...>

🔗jonszanto <jszanto@...>

🔗Mike Battaglia <battaglia01@...>

On Sun, Mar 13, 2011 at 9:52 PM, Carl Lumma <carl@...> wrote:
>
> > This is silly. 435 cents is in 5/4's field of attraction, as you
> > are fond of pointing out. You really require me to test to confirm
> > that the major thirds are within this absurdly large range?
>
> Of *all* intervals **performed**, what fraction are.

There'd better be some reason for why I'm supposed to waste my time
doing this, especially coming from a guy who's fond of saying that the
major thirds in maqam music are 400 cents.

> > > The scales are tetrachordal, which certainly suggests a role
> > > for 3:2. I don't know of any 5:4-period scales among the
> > > maqamat - do you?
> >
> > Which, after I sent out some 5/4-period ajnas, you raised the
> > goalpost and said that although the period here is 400 cents,
>
> Period is not 400 cents. Period is 498 cents. I think
> I said that.

And I said you're wrong. Go back to the site and look at some example
maqams and see how the ajnas fit within one another. Here's an
example:

http://www.maqamworld.com/maqamat/ajam.html#ajam

The Kurd tetrachord starts where the Ajna trichord ends, which is 400
cents up from 0.

I couldn't help but notice that you, once again, ignored my post
asking you to clarify what your precise objections are. This is
roughly the fourth post I've made asking you to clarify, generally
preceded by a paragraph expounding on what it seems like you're
saying. You have ignored every single one of these.

There is no point continuing communicating with you on this further
unless you clarify what it is you're saying, especially if you're fond
of complaining about how I'm misrepresenting your position.

-Mike

🔗Carl Lumma <carl@...>

Mike Battaglia <battaglia01@...> wrote:

> There'd better be some reason for why I'm supposed to waste my
> time doing this, especially coming from a guy who's fond of
> saying that the major thirds in maqam music are 400 cents.

It doesn't get any clearer than how I wrote it the first time.
I suggest you go back and read it.

> > Period is not 400 cents. Period is 498 cents. I think
> > I said that.
>
> And I said you're wrong. Go back to the site and look at
> some example maqams and see how the ajnas fit within one
> another. Here's an example:
>
> http://www.maqamworld.com/maqamat/ajam.html#ajam
>
> The Kurd tetrachord starts where the Ajna trichord ends,
> which is 400 cents up from 0.

All maqam scales, with perhaps a handful of exceptions, are
constructed from tetrachords at the 4:3 or 3:2. This one is
not one of the exceptions.

> There is no point continuing communicating with you on this
> further unless you clarify what it is you're saying, especially
> if you're fond of complaining about how I'm misrepresenting
> your position.

I've written very clearly on this, and every point I have
to make I've made at least twice. I suggest you review this
thread and read what I wrote.

-Carl

🔗Mike Battaglia <battaglia01@...>

On Mon, Mar 14, 2011 at 2:35 AM, Carl Lumma <carl@...> wrote:
>
> Mike Battaglia <battaglia01@...> wrote:
>
> > There'd better be some reason for why I'm supposed to waste my
> > time doing this, especially coming from a guy who's fond of
> > saying that the major thirds in maqam music are 400 cents.
>
> It doesn't get any clearer than how I wrote it the first time.
> I suggest you go back and read it.

I guess I'm just bad at understanding really clear things. Can you
clarify, maybe as though I were a 5 year old?

> > The Kurd tetrachord starts where the Ajna trichord ends,
> > which is 400 cents up from 0.
>
> All maqam scales, with perhaps a handful of exceptions, are
> constructed from tetrachords at the 4:3 or 3:2. This one is
> not one of the exceptions.

*brain explodes*

> > There is no point continuing communicating with you on this
> > further unless you clarify what it is you're saying, especially
> > if you're fond of complaining about how I'm misrepresenting
> > your position.
>
> I've written very clearly on this, and every point I have
> to make I've made at least twice. I suggest you review this
> thread and read what I wrote.

Not only have I reviewed this thread several times now, but at least three
times now I've posted some kind of summary of what I take your
position to be and a request to confirm that it is correct. You've
ignored it every one of those times! A response for one of these would
be helpful in my understanding what exactly you're saying,
particularly when it comes to the psychoacoustic claims that you're
making. This would be helpful for me because I am bad at understanding
really clearly expressed ideas.

-Mike