Yahoo Tuning Groups Ultimate Backup tuning What tuning did this organ have when recorded?

Tom Dent wrote:

> Not a trick question - any method is acceptable to get a good answer.

I was trying to answer a very similar question some time ago. Maybe some frequency analysis software could help but I don't know how. As far as I can hear it, it sounds like meantone with even stronger tempering than 1/4 comma (like 2/7 or so). According to the musical context, I believe that what sounds like G#4 to me is probably the pitch of A4 there, which means the meantone chain goes from Eb to G#. I don't think I can say anything more to it at this time.

Petr

PS: Was the cut end intentional, or has my editor refused to load a portion of the sound?

🔗Tom Dent <stringph@...>

--- In tuning@yahoogroups.com, Petr Parízek <p.parizek@...> wrote:
>
> As far as I can hear it, it sounds like meantone with even stronger
tempering than 1/4 comma (like 2/7 or so). According to the musical
context, I believe that what sounds like G#4 to me is probably the
pitch of A4 there, which means the meantone chain goes from Eb to G#.
I don't think I can say anything more to it at this time.
>
> Petr
>
> PS: Was the cut end intentional, or has my editor refused to load a
portion of the sound?
>

The cut has to come somewhere, as the file in all its wav.y glory is
already 7M long, and the piece (several minutes long) gets rapidly
more complicated. I expect it is much easier to analyze a single tone
than a chord, which is why a fugue beginning with a complete chromatic
scale is a tuning analyst's dream come true.

To deal with possible tape or LP wobble is another question...

There are a couple of websites that claim to specify the tuning, but
I'm not sure how far I trust their cent values, let alone whether it
was touched up between the 'historical' documentation and the
recording. But it ain't stronger than 1/4 comma. Perhaps the slight
blurring of the chromatic scale steps enhances its weirdness.
~~~T~~~

🔗Danny Wier <dawiertx@...>

My guess: 1/3-comma meantone (since chromatic tones sound like third tones), range E flat to G sharp. But I'm using my ear, not a machine.

Or like Petr said, 2/7-comma, or maybe something very similar, like Golden meantone or LucyTuning, who knows... ;)

~D.

Tom Dent wrote:
> http://www.thphys.uni-heidelberg.de/~dent/orgelchrom.wav
>
> Not a trick question - any method is acceptable to get a good answer.
>
> Allowances have to be made for the recording and playback
> technology... including possible LP wobble. I believe some list
> members may have the CD version (Merula played by Leonhardt) which
> would in principle be better.
> ~~~T~~~

🔗Leonardo Perretti <dombedos@...>

Hello Tom and everyone,

yesterday, I had my afternoon with nothing better to do then fiddling with my portable computer, so I decided to take Tom Dent's question as a good exercise for evaluating some sound editing applications I wished to test.
I computed the first 12 frequencies, and the result is the following:

285.4825 Hz.
304.29692
319.47137
340.97939
357.03839
382.18081
400.57803
427.18658
455.27344
477.50181
509.55882
534.67718

I have not Merula's music at hand; anyway, it appears that the first note should be a D, so the eighth would be the A, and the diapason of the organ 427.18658.

The scale in cents (rounded to the second decimal) is:

83.30
196.98
307.48
391.73
504.52
584.20
702.01
783.40
894.74
1004.98
1087.51

With these values, the temperament appears to be a sort of meantone, with the wolf fifth around 1 sc wide, two fifths almost pure (C-G and C#-G#), three fifths of 1/5 or less (F#-C#, Eb-Bb and Bb-F) and the remaining of 1/4 or a bit narrower.
Anyway, everybody can do his evaluation starting from the values above.
I have done all of my best to select the more suitable chunks of sound, but perhaps some frequencies could be refined;

At this point, I am curious to know more about the organ, and also to compare these results with the ones mentioned by Tom.

Should someone be interested, I will describe the method I used.

Best regards to all from a very hot southern Italy.
Leonardo

Tom Dent wrote:
>http://www.thphys.uni-heidelberg.de/~dent/orgelchrom.wav
>
>Not a trick question - any method is acceptable to get a good answer.
>
>Allowances have to be made for the recording and playback
>technology... including possible LP wobble. I believe some list
>members may have the CD version (Merula played by Leonhardt) which
>would in principle be better.
>~~~T~~~

🔗Petr Parízek <p.parizek@...>

Ciao Leonardo,

thank you for the contribution, that's a wonderful job that you've done!
Do you think there's a way to do something similar even if there are no "one-voiced" places? I have a harpsichord recording and I would be very interested to find some more info about the tuning. Unfortunately, I don't think the recording contains any one-voiced passages so I'm not sure if this is possible.

Petr

🔗Tom Dent <stringph@...>

--- In tuning@yahoogroups.com, Leonardo Perretti <dombedos@...> wrote:
>
> The scale in cents (rounded to the second decimal) is:
>
> 83.30
> 196.98
> 307.48
> 391.73
> 504.52
> 584.20
> 702.01
> 783.40
> 894.74
> 1004.98
> 1087.51
>
> With these values, the temperament appears to be a sort of meantone,
> with the wolf fifth around 1 sc wide, two fifths almost pure (C-G and
> C#-G#), three fifths of 1/5 or less (F#-C#, Eb-Bb and Bb-F) and the
> remaining of 1/4 or a bit narrower.
> Anyway, everybody can do his evaluation starting from the values above.
> I have done all of my best to select the more suitable chunks of
> sound, but perhaps some frequencies could be refined;
>
> At this point, I am curious to know more about the organ, and also to
> compare these results with the ones mentioned by Tom.

Looks good... thanks for the work!

Concerning a possible 'LP wobble' problem, it might be interesting to:

1) take the next 3 tones, since the opening chromatic scale goes over
more than an octave (15 notes in total I think), and see how
accurately the octaves come out
2) split each tone into 2, or at least the longer tones, and see
whether there is any significant difference in frequency between the
two halves, this gives you some idea what the uncertainty in the
frequency measurement is.

Since each note lasts about half a second, I think it is unlikely the
frequency determination can be accurate to better than 0.1Hz.

The organ is in the church at Muri (Switzerland), the small
Evangelienorgel built in 1743 by Joseph and Viktor Ferdinand Bossard.
In a publication of Billeter from 1997 its 'historical' tuning is
given as

80.4
195.1
305.9
388.3
502.0
582.4
699.0
779.5
891.2
1001.0
1085.3

however of course there is no guarantee that this was preserved
exactly in the recent restoration: in particular the Bb given here
seems to be too low and was probably altered.

Also, these cent values are consistently *lower* than Leonardo's,
which suggests that his measured 'c' frequency was a little on the low
side. Anyway, the best way to compare is probably to look at the sizes
of the major thirds, since these mainly determine the acoustic effect
of such a tuning. I will hopefully compute these some time today.
~~~T~~~

🔗Tom Dent <stringph@...>

--- In tuning@yahoogroups.com, "Tom Dent" <stringph@...> wrote:
>
> --- In tuning@yahoogroups.com, Leonardo Perretti <dombedos@> wrote:
> >
> > The scale in cents (rounded to the second decimal) is:
> >
> > 83.30
> > 196.98
> > 307.48
> > 391.73
> > 504.52
> > 584.20
> > 702.01
> > 783.40
> > 894.74
> > 1004.98
> > 1087.51
> >
>
> The organ is in the church at Muri (Switzerland), the small
> Evangelienorgel built in 1743 by Joseph and Viktor Ferdinand Bossard.
> In a publication of Billeter from 1997 its 'historical' tuning is
> given as
>
> 80.4
> 195.1
> 305.9
> 388.3
> 502.0
> 582.4
> 699.0
> 779.5
> 891.2
> 1001.0
> 1085.3
>
> Anyway, the best way to compare is probably to look at the sizes
> of the major thirds, since these mainly determine the acoustic effect
> of such a tuning. I will hopefully compute these some time today.

Here are the major third values, first from Leonardo's measurement of
the first 12 notes:

c 391.7 e 391.7 g# 416.6
g 385.5 b 420.0 eb 394.5
d 387.2 f# 420.8 bb 392
a 388.6 c# 421.2 f 390.2

second from the organological publication

c 388.3 e 391.2 g# 420.5
g 386.3 b 420.5 eb 393.2
d 387.3 f# 418.6 bb 394.1
a 389.2 c# 421.5 f 389.2

This backs up the guesses that 1) Leonardo's c was a bit too low
relative to the actual value as played 2) the published value of bb
was a bit too low relative to general musical expectations, and also
suggests 3) Leonardo's g was a bit too high. However, we cannot really
tell what the sources of such discrepancies are. (And, of course, what
the discrepancy over 250 years from the real historical tuning was...)

Apart from these notes we have several excellent matches with
deviations of +0.5, -0.5, -0.1, -0.6, -0.3, +1 for the thirds e-g#,
b-eb, d-f#, a-c#, c#-f, f-a.

In any case we have a slightly modified meantone tuning with some pure
thirds, g-c-f-bb-eb are tempered slightly less than 1/4 comma as are
f#-c#-g#, thus the wolf g#-eb is about 1 comma wide rather than >= 3/2
comma as in more strict meantone.
~~~T~~~

🔗Leonardo Perretti <dombedos@...>

Thank you Petr and Tom for your replies.

Petr wrote:
>Do you think there's a way to do something similar even if there are >no "one-voiced" places? I have a harpsichord recording and I would >be very interested to find some more info about the tuning. >Unfortunately, I don't think the recording contains any one-voiced >passages so I'm not sure if this is possible.

The method I used here is a "semi-artisanal" one; it is very accurate, but it requires that the operator can identify and mark "by hand" the zero-crossing points directly on the waveshape. If the sound is complex, overlapping of the sounds from the single strings prevents from identifyng the zero-crossing points of the single waveshapes; this is why it is usable only with one-note sounds. The case of harpsichord is even more complicated by the fact that decay of the strings' sound is fast, expecially in the treble section, so the possible accuracy is reduced.
Maybe it is possible to obtain purer, suitable sounds by applying sound filtering with very narrow parameters, but I never tried it, so I cannot say.
If you like, please send me the recording, or part of it, so I can give it a look.

Tom wrote:
>Concerning a possible 'LP wobble' problem, it might be interesting to:
>
>1) take the next 3 tones, since the opening chromatic scale goes over
>more than an octave (15 notes in total I think), and see how
>accurately the octaves come out
>2) split each tone into 2, or at least the longer tones, and see
>whether there is any significant difference in frequency between the
>two halves, this gives you some idea what the uncertainty in the
>frequency measurement is.

On next Sunday, I have some time to dedicate, hopefully; I'll try it.

>Since each note lasts about half a second, I think it is unlikely the
>frequency determination can be accurate to better than 0.1Hz.

As far as I can say, accuracy should be around 0.05 or so; take into account that I used only a fraction (more or less 1/3) of the notes for the calculation. I will evaluate the actual level of exactness.
Also, I will try to evaluate the effect of LP wobble, and verify the discrepancies with respect to Billeter data, if time permits.

🔗Petr Parízek <p.parizek@...>

🔗Tom Dent <stringph@...>

--- In tuning@yahoogroups.com, Petr Parízek <p.parizek@...> wrote:
>
> To Leonardo:
>
> I've made an external link to the harpsichord recording. It should
stay there for one week. Here it is:
www.yousendit.com/download/Q01FT2pIcVh6NEkwTVE9PQ
>
> Petr
>

I suspect this will be much more difficult to evaluate scientifically
as 1) it consists almost exclusively of full chords or fast moving
passages 2) the pitch of a harpsichord is subject to an initial
transient, hence one cannot really use the very short notes, at least
one has to be very careful 3) the sound is mostly 'dry' and notes fade
away very quickly.

However, the passage from about 3'30'' to 4'45'' is played with only
one register and has a lot more sustain (due to sympathetic resonance
of the other register being undamped) so one can perhaps get some
better idea of pitches from that.

Simply by listening and by knowing what is likely to happen in the
'early-music' scene, I would bet quite strongly on quarter-comma
meantone.

Who is playing?
~~~T~~~

🔗Petr Parízek <p.parizek@...>

🔗Leonardo Perretti <dombedos@...>

Thank you Petr; I have downloaded and listened to it.

I fully second Tom's opinions; anyway, I will try
something when I have a bit of time.

Leonardo

At 22:17 +0200 7-08-2008, Petr Parízek wrote:
>To Leonardo:
>
>I've made an external link to the harpsichord
>recording. It should stay there for one week.
>Here it is:
><http://www.yousendit.com/download/Q01FT2pIcVh6NEkwTVE9PQ>www.yousendit.com/download/Q01FT2pIcVh6NEkwTVE9PQ
>
>Petr

At 20:45 +0000 7-08-2008, Tom Dent wrote:
>I suspect this will be much more difficult to evaluate scientifically
>as 1) it consists almost exclusively of full chords or fast moving
>passages 2) the pitch of a harpsichord is subject to an initial
>transient, hence one cannot really use the very short notes, at least
>one has to be very careful 3) the sound is mostly 'dry' and notes fade
>away very quickly.
>
>However, the passage from about 3'30'' to 4'45'' is played with only
>one register and has a lot more sustain (due to sympathetic resonance
>of the other register being undamped) so one can perhaps get some
>better idea of pitches from that.

🔗Carl Lumma <carl@...>

🔗robert thomas martin <robertthomasmartin@...>

--- In tuning@yahoogroups.com, Leonardo Perretti <dombedos@...> wrote:
>
> Hello Tom and everyone,
>
> yesterday, I had my afternoon with nothing better to do then
fiddling
> with my portable computer, so I decided to take Tom Dent's question
> as a good exercise for evaluating some sound editing applications I
> wished to test.
> I computed the first 12 frequencies, and the result is the
following:
>
> 285.4825 Hz.
> 304.29692
> 319.47137
> 340.97939
> 357.03839
> 382.18081
> 400.57803
> 427.18658
> 455.27344
> 477.50181
> 509.55882
> 534.67718
>
> I have not Merula's music at hand; anyway, it appears that the
first
> note should be a D, so the eighth would be the A, and the diapason
of
> the organ 427.18658.
>
> The scale in cents (rounded to the second decimal) is:
>
> 83.30
> 196.98
> 307.48
> 391.73
> 504.52
> 584.20
> 702.01
> 783.40
> 894.74
> 1004.98
> 1087.51
>
> With these values, the temperament appears to be a sort of
meantone,
> with the wolf fifth around 1 sc wide, two fifths almost pure (C-G
and
> C#-G#), three fifths of 1/5 or less (F#-C#, Eb-Bb and Bb-F) and the
> remaining of 1/4 or a bit narrower.
> Anyway, everybody can do his evaluation starting from the values
above.
> I have done all of my best to select the more suitable chunks of
> sound, but perhaps some frequencies could be refined;
>
> At this point, I am curious to know more about the organ, and also
to
> compare these results with the ones mentioned by Tom.
>
> Should someone be interested, I will describe the method I used.
>
>
> Best regards to all from a very hot southern Italy.
> Leonardo
>
>
>
> Tom Dent wrote:
> >http://www.thphys.uni-heidelberg.de/~dent/orgelchrom.wav
> >
> >Not a trick question - any method is acceptable to get a good
answer.
> >
> >Allowances have to be made for the recording and playback
> >technology... including possible LP wobble. I believe some list
> >members may have the CD version (Merula played by Leonhardt) which
> >would in principle be better.
> >~~~T~~~
>

From Robert. Your methodology is very interesting. Could you please
explain more about the software that you use and how budding
musicologists might use your methods to investigate mp3 files
which are freely available on the internet. Any software which
gives an accuracy of plus or minus 6cents has to be given serious
consideration.