Topic by JohnSchenker

  • Avatar
  • JohnSchenker
  • Member
  • Offline
  • Registered: 17-02-2016
  • Posts: 7

Topic: Tempest great tool for estimating the temperament from recording

Surprisingly I have not found mentioned one of the greatest research tools for investigating temperaments:

TempEst

http://www.isophonics.net/sawa/tempest/

just upload an .mp3 or .wav file with a recorded solo piece played on a harpsichord with no other instruments, (YES harpsichord ONLY, NOT PIANO)



I am still trying to figure out how it works, according to this research paper, in its development they used the modeled Harpsichords from Pianoteq


The real harpsichord recordings were played by Dan
Tidhar on a Rubio double-manual harpsichord in a small
hall. The synthesized recordings were performed on a digital
keyboard by Dan Tidhar and rendered from MIDI files using
the physical modeling synthesis software Pianoteq (Pianoteq,
2010).

the version used was Pianoteq 3 see the acknowledgements section of Estimation of harpsichord inharmonicity and temperament from musical recordings


[h1]Trying out TempEst the Online tool for estimatings instrument tuning parameters[/h1]

But when trying to identify temperaments, I created a recording of the piece F.Couperin "Les baricades misterieuses"  (Download midi file here using the scala file “werck3.scl” copied below:


! werck3.scl
!
Andreas Werckmeister's temperament III (the most famous one, 1681)
 12
!
 256/243
 192.18000
 32/27
 390.22500
 4/3
 1024/729
 696.09000
 128/81
 888.26999
 16/9
 1092.18000
 2/1

which can be downloaded here
http://www.huygens-fokker.org/docs/scales.zip
werck3.scl “Andreas Werckmeister's temperament III (the most famous one, 1681)”


and it was labeled as 

    rdfs:label """SixthCommaMeantone_shift""" ;


and the closest sounding temperament I have been able to find on YouTube to this werck3.scl (with a=415) is this Elena Zhukova Video here:

Elena Zhukova

which was labeled:

    rdfs:label """Valotti_shift""" ;

I then recorded the audio of the Youtube video version by claudio di Veroli:


Claudio di Veroli (Tuning: ordinaire (Rousseau) temperament)


and TempEst labeled it as


 rdfs:label """FifthComma""" ;

I found This other temperament in Buxtehude - Toccata in G - BuxWV 164- Christian Zell Harpsichord (Werckmeister III) also descrbed as a:


    rdfs:label """FifthComma""" ;
Last edited by JohnSchenker (17-02-2016 16:49)

Reply by SteveLy

  • Avatar
  • SteveLy
  • Member
  • Offline
  • From: Melbourne, Australia
  • Registered: 16-11-2015
  • Posts: 377

Re: Tempest great tool for estimating the temperament from recording

I would not rely on youtube videos for precise pitch comparisons. Video format conversions often slightly alter frame rates which will shift the pitch (occasionally a lot, by up to about a semitone; commonly used frame rates for different formats: 24, 25, 29.97, 30, 50, 59.94, 60 fps).

Last edited by SteveLy (17-02-2016 19:26)
3/2 = 5

SteveLy's Website

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,342

Re: Tempest great tool for estimating the temperament from recording

I tried TempEst using a short test with a pianoteq harpsichord tuned at the built-in Vallotti Scala tuning and the software got it right (but spells it wrong...) :

:temperament_0 a tm:Temperament ,
         tm:Valotti ;
     tm:description :description_00 ,
         :description_01 .


I then tried it with this piece that I have on CD : Jean Rondeau - Lute Suite No 2 In C Minor, Bwv 997 I Prelude
I like the tuning but it is not specified on the cover. The harpsichord is set at about A=411 Hz in this piece.

The result was:

:temperament_0 a tm:SixthCommaMeantone ,
         tm:Temperament ;
     tm:description :description_00 ,
         :description_01 .

This tuning is available also for pianoteq and it sounds quite plausible when playing a pianoteq harpsichord.

A useful tool indeed...

Last edited by Gilles (17-02-2016 21:56)

Reply by JohnSchenker

  • Avatar
  • JohnSchenker
  • Member
  • Offline
  • Registered: 17-02-2016
  • Posts: 7

Re: Tempest great tool for estimating the temperament from recording

Gilles wrote:

...

I then tried it with this piece that I have on CD : Jean Rondeau - Lute Suite No 2 In C Minor, Bwv 997 I Prelude
I like the tuning but it is not specified on the cover. The harpsichord is set at about A=411 Hz in this piece.

The result was:

:temperament_0 a tm:SixthCommaMeantone ,
         tm:Temperament ;
     tm:description :description_00 ,
         :description_01 .

You did probably spot it but... just in case...TempEst comes with a=415 as default, but it can be changed with the Configure Tab


TempEst_Configure_Tab

Last edited by JohnSchenker (18-02-2016 14:45)

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,342

Re: Tempest great tool for estimating the temperament from recording

JohnSchenker wrote:
Gilles wrote:

...

I then tried it with this piece that I have on CD : Jean Rondeau - Lute Suite No 2 In C Minor, Bwv 997 I Prelude
I like the tuning but it is not specified on the cover. The harpsichord is set at about A=411 Hz in this piece.

The result was:

:temperament_0 a tm:SixthCommaMeantone ,
         tm:Temperament ;
     tm:description :description_00 ,
         :description_01 .

You did probably spot it but... just in case...TempEst comes with a=415 as default, but it can be changed with the Configure Tab

No, I had not looked at this tab, I thought the diapason was automatically discovered...In any case it does make a difference since rerunning it with A=411Hz returned Vallotti as the correct tuning! I'm not surprised because afterwards, trying to play the prelude along with the recording, I found the Ab too flat with SixthCommaMeantone while Vallotti sounded much better. It is also known that the makers of Rondeau's harpsichord (Jonte Knif & Arno Pelto) seem to favor Vallotti in other circumstances than this CD, so no surprise really.

I must also say that I added the A=411Hz comment afterwards since at first I thought the standard A=415Hz was used. It's only while playing along with the piece that I noticed the difference. I can't upload the Rondeau example since it is a commercial recording, but you can sample it online here: http://www.deezer.com/artist/5368041

Last edited by Gilles (18-02-2016 15:30)

Reply by JohnSchenker

  • Avatar
  • JohnSchenker
  • Member
  • Offline
  • Registered: 17-02-2016
  • Posts: 7

Re: Tempest great tool for estimating the temperament from recording

No, I had not looked at this tab, I thought the diapason was automatically discovered...In any case it does make a difference since rerunning it with A=411Hz returned Vallotti as the correct tuning! I'm not surprised because afterwards, trying to play the prelude along with the recording, I found the Ab too flat with SixthCommaMeantone while Vallotti sounded much better. It is also known that the makers of Rondeau's harpsichord (Jonte Knif & Arno Pelto) seem to favor Vallotti in other circumstances than this CD, so no surprise really.

I must also say that I added the A=411Hz comment afterwards since at first I thought the standard A=415Hz was used. It's only while playing along with the piece that I noticed the difference. I can't upload the Rondeau example since it is a commercial recording, but you can sample it online here: http://www.deezer.com/artist/5368041


Thank you very much for the link.


I think I have found my answer....

I was trying different temperaments in Pianoteq  (version 4.5.5) so when i used the built-in Werkmeister III Temperament in Grimaldi A it was playing in a=415 but when I  switched to microtuning mode to upload the  Scala file "werck3.scl, the default keymap overruns this a=415 with the default 440 specified in the .kbm file so it plays with the Werkmeister III tuned to a=440, as I understood from this post by Philippe Guillaume

The default keymap will map the first note (omitted in scala format) on middle C and put the A at 440Hz (thus the 12 notes above range from midi 61 (C#3) to 72 (C4)).
For other mappings, you will need to load also a kbm file, see for example

Last edited by JohnSchenker (18-02-2016 18:48)

Reply by JohnSchenker

  • Avatar
  • JohnSchenker
  • Member
  • Offline
  • Registered: 17-02-2016
  • Posts: 7

Re: Tempest great tool for estimating the temperament from recording

Even though the results of TempEst are very accurate and I believe this online software to be the key to study temperaments, I still cannot get this online software to identify equal temperament from a .wav or .mp3 file of a midi rendered by Pianoteq, either playing the built-in Equal temperament, or the neidhardt4.scl Equal Temperament scala file. Surely if they had 100% success (read it here) identifying the temperament of the midis rendered by Pianoteq Version 3 My 4.5.5 version should be as accurate.

Last edited by JohnSchenker (18-02-2016 19:06)

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,342

Re: Tempest great tool for estimating the temperament from recording

Not sure if this old post by Philippe is still true. The default keymap is linear, and if you set first the diapason, then load a scala file the diapason is maintained correctly.
I tried the werck3.scl with first setting A=415 and it stays at A=415. You can change it afterwards also without problem. I'm using version 5.5.1.

Last edited by Gilles (18-02-2016 19:01)

Reply by JohnSchenker

  • Avatar
  • JohnSchenker
  • Member
  • Offline
  • Registered: 17-02-2016
  • Posts: 7

Re: Tempest great tool for estimating the temperament from recording

Gilles wrote:

Not sure if this old post by Philippe is still true. The default keymap is linear, and if you set first the diapason, then load a scala file the diapason is maintained correctly.
I tried the werck3.scl with first setting A=415 and it stays at A=415. You can change it afterwards also without problem. I'm using version 5.5.1.



As explained above the version that I have 4.5.5 behaves like this when I  switched to microtuning mode to upload the  Scala file "werck3.scl, the default keymap overruns this a=415 with the default 440 specified in the .kbm file so it plays with the Werkmeister III tuned to a=440

Last edited by JohnSchenker (18-02-2016 19:29)