Topic by bm

  • Avatar
  • bm
  • Member
  • Offline
  • From: France (Burgundy)
  • Registered: 15-09-2017
  • Posts: 302

Topic: Pianoteq spectrum profile

If tonal equalization downstream of the Pianoteq virtual sound generator is not necessarily critical to have a (subjective) impression of a realistic sound, management of the spectrum profile upstream - note by note (Pianoteq pro) seems to me very important to adjust the tone of the instrument.

There are currently several limitations of spectrum profile note by note tool that, perhaps ? may be worth it an evolution in Pianoteq pro: (provided that it is not too heavy for the CPUs - or maybe it can be pre-calculated (pre-compilation)?)

1) the spectrum profile of the resonance of the piano case in low frequencies, induced by (sorry for the blur of this amateur reduction) keyboard mecanism and hammers.
When a real piano is recorded note by note, and we look at its spectrum profile (for example with Audacity), we immediately notice peaks at frequencies lower than frequency of the fundamental harmonic of the note.
2) the spectrum profile note tool by current pianoteq note appears? to be taken into account MAINLY for notes played loudly. (ff/fff) and at the beginning of the attack-decay phase.
Maybe it should?, be able to set the spectrum profile separately for the 3 levels (piano, mezzo, forte), as it is proposed for these 3 levels and note by note for Hammer Hardness.

Can Pianoteq's application of its mathematical model be in the future parameterized also on these criteria? (without requiring an extremely powerful computer).
I sometimes have the impression (purely personal opinion perhaps not shared) that the sound at the piano and mezzo level is sometimes a little too deaf compared to a real piano.
The above suggestion may not be, perhaps ? the right approach to improve realism on this point. (Should we act with the current version of Pianoteq rather by combining a different setting of the hardness of the piano and mezzo hammers and the velocity curve of the keyboard ?, or other tuning ?)

Last edited by bm (11-07-2018 17:56)

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,345

Re: Pianoteq spectrum profile

bm wrote:

1) the spectrum profile of the resonance of the piano case in low frequencies, induced by (sorry for the blur of this amateur reduction) keyboard mecanism and hammers.
When a real piano is recorded note by note, and we look at its spectrum profile (for example with Audacity), we immediately notice peaks at frequencies lower than frequency of the fundamental harmonic of the note.

Maybe using for reverb this impulse response made from knocking on a piano case can help:

knock.wav

available in https://sites.google.com/site/ptqspecprof/temp

It can also be complemented with a room reverb (by simple wav addition) such as this pre-calculated example:

Wide Echo Hall+Knock.wav

Last edited by Gilles (11-07-2018 19:57)

Reply by dklein

  • Avatar
  • dklein
  • Member
  • Offline
  • From: Florida
  • Registered: 29-11-2015
  • Posts: 1,600

Re: Pianoteq spectrum profile

Used knock.wav in the IR convolution engine in Pianoteq, and like it for "case reverb".

- David

Reply by bm

  • Avatar
  • bm
  • Member
  • Offline
  • From: France (Burgundy)
  • Registered: 15-09-2017
  • Posts: 302

Re: Pianoteq spectrum profile

Gilles wrote:
bm wrote:

1) the spectrum profile of the resonance of the piano case in low frequencies, induced by (sorry for the blur of this amateur reduction) keyboard mecanism and hammers.
When a real piano is recorded note by note, and we look at its spectrum profile (for example with Audacity), we immediately notice peaks at frequencies lower than frequency of the fundamental harmonic of the note.

Maybe using for reverb this impulse response made from knocking on a piano case can help:

knock.wav

available in https://sites.google.com/site/ptqspecprof/temp

Thank you Gille
to have made us discover this very interesting function (effects-> reverb-> type: Impulse Wav). It seems to make it possible to improve - in part - the taking into account of the coloration of the sound by the case. (tested with your knock.wav file and the Bluethner One BA preset - after adjusting the level of this effect, including the duration of the sound envelope)
Question about your knock file: From what real piano model was this file recorded?
For the Pianoteq team: I have the impression (at least with the Bluethner preset) that the duration of the IR effect (possible bug?) Is abnormally long just with 2 notes: C1 and C2 (midi 36 and 48) while the duration is normal with all other note

Regards

Bruno

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,345

Re: Pianoteq spectrum profile

bm wrote:

Question about your knock file: From what real piano model was this file recorded?

Regards

Bruno

I don't remember where the file was found, a couple of years ago during a beta session by someone else, so I had kept it. The name being too general, I couldn't find it on the internet since files with this name are mostly door knock. The piano was probably a large upright. Looking again today, I found this one that sounds interesting:

https://freesound.org/people/patchen/sounds/8928/

Seems you have to login to download it though.

This one seems better because there is no harp sound from undamped strings, the case is empty. I suppose the one I keep was made to be like a spring reverb more than to emulate the cabinet.

The idea of adding a separate reverb to it is mine though...

Reply by bm

  • Avatar
  • bm
  • Member
  • Offline
  • From: France (Burgundy)
  • Registered: 15-09-2017
  • Posts: 302

Re: Pianoteq spectrum profile

Gilles wrote:
bm wrote:

Question about your knock file: From what real piano model was this file recorded?

Regards

Bruno

I don't remember where the file was found, a couple of years ago during a beta session by someone else, so I had kept it. The name being too general, I couldn't find it on the internet since files with this name are mostly door knock. The piano was probably a large upright. Looking again today, I found this one that sounds interesting:

https://freesound.org/people/patchen/sounds/8928/

Seems you have to login to download it though.

This one seems better because there is no harp sound from undamped strings, the case is empty. I suppose the one I keep was made to be like a spring reverb more than to emulate the cabinet.

The idea of adding a separate reverb to it is mine though...

The problem is that we do not know which piano case (It sounds differently from one type of piano to another) it is and it should be possible (ideally) also tune the spectrum profile of the case note by note.
The goal here is not to manage reverb.
Regards,

Bruno

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,345

Re: Pianoteq spectrum profile

bm wrote:
Gilles wrote:
bm wrote:

Question about your knock file: From what real piano model was this file recorded?

Regards

Bruno

I don't remember where the file was found, a couple of years ago during a beta session by someone else, so I had kept it. The name being too general, I couldn't find it on the internet since files with this name are mostly door knock. The piano was probably a large upright. Looking again today, I found this one that sounds interesting:

https://freesound.org/people/patchen/sounds/8928/

Seems you have to login to download it though.

This one seems better because there is no harp sound from undamped strings, the case is empty. I suppose the one I keep was made to be like a spring reverb more than to emulate the cabinet.

The idea of adding a separate reverb to it is mine though...

The problem is that we do not know which piano case (It sounds differently from one type of piano to another) it is and it should be possible (ideally) also tune the spectrum profile of the case note by note.
The goal here is not to manage reverb.
Regards,

Bruno

Since you have a piano and mics, maybe you can record your own...
By the way, there is some case sound in the Pianoteq instruments, it's just not very obvious but can be seen in Audacity spectra, I think.

Last edited by Gilles (12-07-2018 16:46)

Reply by bm

  • Avatar
  • bm
  • Member
  • Offline
  • From: France (Burgundy)
  • Registered: 15-09-2017
  • Posts: 302

Re: Pianoteq spectrum profile

Gilles wrote:

Since you have a piano and mics, maybe you can record your own...
By the way, there is some case sound in the Pianoteq instruments, it's just not very obvious but can be seen in Audacity spectra, I think.

If I make my own recording of the piano case, and then I integrate it in the form of IR (via reverb menu) I imagine that this will be problematic on several points:
1 there will be interference with the sound of the piano case already recorded with Pianoteq's piano model.
2 it does not seem possible in current pianoteq pro to use (note by note) an equalizer function on the only coloration of the sound provided by the piano case independently of the possible overall equalization downstream of the sound generator, or independently of adjusting the balance between the different partials upstream of the sound generator (spectrum profile). - I'm sorry, I have a hard time expressing this idea - it should be "something" comparable to the spectrum profile function, maybe a simple equalizer because on frequencies unrelated simple with the harmonics of a vibrating string , dedicated to the adjustment of the only coloration of the sound by the piano case.
Maybe we have to use other functions of Pianoteq to achieve this result?

Bruno

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,345

Re: Pianoteq spectrum profile

bm wrote:
Gilles wrote:

Since you have a piano and mics, maybe you can record your own...
By the way, there is some case sound in the Pianoteq instruments, it's just not very obvious but can be seen in Audacity spectra, I think.

If I make my own recording of the piano case, and then I integrate it in the form of IR (via reverb menu) I imagine that this will be problematic on several points:
1 there will be interference with the sound of the piano case already recorded with Pianoteq's piano model.
2 it does not seem possible in current pianoteq pro to use (note by note) an equalizer function on the only coloration of the sound provided by the piano case independently of the possible overall equalization downstream of the sound generator, or independently of adjusting the balance between the different partials upstream of the sound generator (spectrum profile). - I'm sorry, I have a hard time expressing this idea - it should be "something" comparable to the spectrum profile function, maybe a simple equalizer because on frequencies unrelated simple with the harmonics of a vibrating string , dedicated to the adjustment of the only coloration of the sound by the piano case.
Maybe we have to use other functions of Pianoteq to achieve this result?

Bruno

I understand your need to overcome the limits of Pianoteq Pro but for the time being we have to live with them. The use of a case IR as reverb is of course a trick, something more like an effect than a modeling option. I'm pretty sure what you want is complete control over the modelization process, something only Modartt has access to at the time. The reason is surely that for real-time operation, a lot of things have to be pre-calculated and we only have access to parameters that have a small effect on the real-time performance.

If I recall well, Philippe once said that thousands of parameters were needed for the modeling process and making a proper GUI for that would be almost impossible.

Reply by bm

  • Avatar
  • bm
  • Member
  • Offline
  • From: France (Burgundy)
  • Registered: 15-09-2017
  • Posts: 302

Re: Pianoteq spectrum profile

Gilles wrote:

I understand your need to overcome the limits of Pianoteq Pro but for the time being we have to live with them. The use of a case IR as reverb is of course a trick, something more like an effect than a modeling option. I'm pretty sure what you want is complete control over the modelization process, something only Modartt has access to at the time. The reason is surely that for real-time operation, a lot of things have to be pre-calculated and we only have access to parameters that have a small effect on the real-time performance.

If I recall well, Philippe once said that thousands of parameters were needed for the modeling process and making a proper GUI for that would be almost impossible.

Finally, it is not so bad.
There is already a lot to do (and learn) with the existing settings of Pianoteq Pro and an already excellent sound generator that continues to improve.
Setting up the case with IR from its own recording will only require not hitting too hard to keep the actual instruments healthy ...
Thanks again for these tips, and probably see you soon for more questions on this exciting topic. (He's just getting a little late in France)

Regards,

Bruno

Reply by _DJ_

  • Avatar
  • _DJ_
  • Member
  • Offline
  • Registered: 01-10-2015
  • Posts: 251

Re: Pianoteq spectrum profile

Gilles wrote:

The use of a case IR as reverb is of course a trick, something more like an effect than a modeling option.

This was recently discussed (somewhat heatedly and among many other topics) here.  To summarize: in the instance of my setup & custom fxps under 5.x, I used to use various IR's for this very trick of "simulating case resonance", however user ModellingAudioProd points out (and I subsequently corroborated) that this is really no longer a benefit of sorts under 6.x (if one is interested in perceived realism of the sound) as the model already accounts for this.  Adding reverb to/in place of resonance only obfuscates the sound... this may certainly be a pleasing obfuscation, but seems to be at a removal from the sound of "the instrument in the room".
cheers,
dj

Matthieu 7:6

Reply by bm

  • Avatar
  • bm
  • Member
  • Offline
  • From: France (Burgundy)
  • Registered: 15-09-2017
  • Posts: 302

Re: Pianoteq spectrum profile

_DJ_ wrote:
Gilles wrote:

The use of a case IR as reverb is of course a trick, something more like an effect than a modeling option.

This was recently discussed (somewhat heatedly and among many other topics) here.  To summarize: in the instance of my setup & custom fxps under 5.x, I used to use various IR's for this very trick of "simulating case resonance", however user ModellingAudioProd points out (and I subsequently corroborated) that this is really no longer a benefit of sorts under 6.x (if one is interested in perceived realism of the sound) as the model already accounts for this.  Adding reverb to/in place of resonance only obfuscates the sound... this may certainly be a pleasing obfuscation, but seems to be at a removal from the sound of "the instrument in the room".
cheers,
dj

Thank you for your feedback and this link.
This joins my first impressions: a less natural sound when using IR (trick) - at the expense of the reverb - to get closer to the coloration of the case. So no need to mistreat my instruments ... with the current version of Pianoteq pro.

Regards,

Bruno

Last edited by bm (13-07-2018 08:48)

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,345

Re: Pianoteq spectrum profile

bm wrote:
_DJ_ wrote:
Gilles wrote:

The use of a case IR as reverb is of course a trick, something more like an effect than a modeling option.

This was recently discussed (somewhat heatedly and among many other topics) here.  To summarize: in the instance of my setup & custom fxps under 5.x, I used to use various IR's for this very trick of "simulating case resonance", however user ModellingAudioProd points out (and I subsequently corroborated) that this is really no longer a benefit of sorts under 6.x (if one is interested in perceived realism of the sound) as the model already accounts for this.  Adding reverb to/in place of resonance only obfuscates the sound... this may certainly be a pleasing obfuscation, but seems to be at a removal from the sound of "the instrument in the room".
cheers,
dj

Thank you for your feedback and this link.
This joins my first impressions: a less natural sound when using IR (trick) - at the expense of the reverb - to get closer to the coloration of the case. So no need to mistreat my instruments ... with the current version of Pianoteq pro.

Regards,

Bruno

Ah, yes, I surely agree that the "trick" is not necessary any more...and I sure didn't want you to hurt your piano by hitting it!

The use of the reverb as an added effect goes way back to version 4.0 circa 2012 where the most common complaint was the "lack of wood" in Pianoteq's pianos. This is not common any more due to the advances in modelling. The addition of the convolution reverb (in version 4.0.0) gave the idea to experiment with it. I even used it to simulate a guitar from the Neupert clavichord by using the knock.wav IR to create a guitar body. Here is the 2012 demo (with version 4):

https://www.forum-pianoteq.com/uploads....Vendra.mp3

For a purist, nothing should be added to the instrument modeled, I agree, but it is to be noted that most version 6 instruments all contain some effects: delay to subliminaly thicken the sound, equalization to sometimes correct the micing side-effects. Some reverb is always present in the stock presets even though playing live in a room or hall through speakers should use no reverb. All these are user selectable options, contrary to samples where the recorded reverb is always there...

Last edited by Gilles (13-07-2018 14:13)

Reply by _DJ_

  • Avatar
  • _DJ_
  • Member
  • Offline
  • Registered: 01-10-2015
  • Posts: 251

Re: Pianoteq spectrum profile

Gilles wrote:

Here is the 2012 demo (with version 4):

https://www.forum-pianoteq.com/uploads....Vendra.mp3

this is bloody awesome! 
i love it as it is, but have you given any thought to remastering it on 6.2.1?

Matthieu 7:6

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,345

Re: Pianoteq spectrum profile

_DJ_ wrote:
Gilles wrote:

Here is the 2012 demo (with version 4):

https://www.forum-pianoteq.com/uploads....Vendra.mp3

this is bloody awesome! 
i love it as it is, but have you given any thought to remastering it on 6.2.1?

Unfortunately, the version 4 of the clavichord had a different name and the fxp can't be loaded in version 6, but here is an example with the stock Neupert Lute preset with the knock.wav added and pushed to +6 dB. In this  short example (first minute or so), I switch the reverb on and off at times so the added "body" can be clearly heard.

https://www.forum-pianoteq.com/uploads....%20duo.mp3

(Sorry, the external recording software adds a bit of silence at the beginning (and I was too lazy to edit it...))

There is a clear difference in overall tone with the version 4 demo, and it is probably due to the advances in modelling and not the old fxp that I can't reuse.

Last edited by Gilles (13-07-2018 15:42)