Topic by lowendtheory

  • Avatar
  • lowendtheory
  • Member
  • Offline
  • Registered: 28-12-2012
  • Posts: 97

Topic: Redoing string re-excitation to use only 1 polyphony

While looking at the performance monitor, I noticed that a couple of shortcuts were taken to allow the impression of round-robins. Each time a string is re-struck, the polyphony counter increments by 1, implying the behavior to be closer to a ROMpler in repeating samples of the same note on top of each other, even though one would expect that not to be the case considering it's a physically-based model. Can you go into more detail concerning the string excitation model as it currently stands? Hopefully when the software moves into later versions, the excitation will be re-calculated for each successive strike, keeping the polyphony per note at one, as in a real piano, instead of the current behavior.

Reply by delt

  • Avatar
  • delt
  • Member
  • Offline
  • From: Montréal, Canada
  • Registered: 11-06-2013
  • Posts: 401

Re: Redoing string re-excitation to use only 1 polyphony

I also noticed this and am curious about it.

http://soundcloud.com/delt01
Pianoteq 5 STD+blüthner, Renoise 3 • Roland FP-4F + M-Audio Keystation 88es
Intel i5@3.4GHz, 16GB • Linux Mint xfce 64bit

Reply by EvilDragon

  • Avatar
  • EvilDragon
  • Member
  • Offline
  • From: Croatia
  • Registered: 27-06-2008
  • Posts: 3,994

Re: Redoing string re-excitation to use only 1 polyphony

The current note does interact the previous note, even if the same key is played, that's why it takes another voice of polyphony. Even on a real piano the previous note is not just cut off...

Hard work and guts!

Reply by User123123123

  • Avatar
  • User123123123
  • Member
  • Offline
  • Registered: 16-09-2013
  • Posts: 63

Re: Redoing string re-excitation to use only 1 polyphony

The same three strings are vibrating. If Pianoteq really models strings, then they do NOT need 2 voices for a repeated keystroke. Of course the vibration modes of the strings will be a little different when triggering a string that already vibrates comparing to triggering a string that is quiet. But if good modeling is done, this should not be simulated using multiple voices. That has to be taken care of in the modeling of the string vibrations.

Reply by EvilDragon

  • Avatar
  • EvilDragon
  • Member
  • Offline
  • From: Croatia
  • Registered: 27-06-2008
  • Posts: 3,994

Re: Redoing string re-excitation to use only 1 polyphony

I think Phillippe would be the best to answer this, but I'm quite positive that multiple voices are very much needed even if the same set of strings is re-excited.

Hard work and guts!

Reply by skip

  • Avatar
  • skip
  • Member
  • Offline
  • From: Sydney, Australia
  • Registered: 21-12-2007
  • Posts: 1,059

Re: Redoing string re-excitation to use only 1 polyphony

I noticed this too, and I too assume that it is doing the same as a ROMpler, but it's hard to be absolutely sure. 

Regarding  a real piano's strings not cutting off, well, I disagree. They are "cut off" at the instant the hammer strikes them again, and re-excited. (in interesting varying ways, due to the fact that the hammer is striking strings that are already in motion)   If Pianoteq is overlapping voices, that most certainly is not how a real piano works.

If Pianoteq is overlapping, then perhaps, if it one day does model it instead of overlapping, the sound clarity might improve a bit for repeated notes with the sustain pedal. Just a thought. (not that I really notice anything wrong with the clarity as it is now, though)

Greg.

Last edited by skip (25-11-2013 10:04)

Reply by julien

  • Avatar
  • julien
  • Admin
  • Offline
  • Registered: 19-04-2007
  • Posts: 1,026

Re: Redoing string re-excitation to use only 1 polyphony

Well, one could argue that there should be no voice management at all as all strings are always vibrating by sympathy. Voice management here is just a mean to reduce, and control, cpu usage. There are extra computations that must take place each time a hammer strikes the strings.

Repeated notes do not sound the same, you can hear how the longitudinal modes build up in the bass, or how phases cancel sometimes themselves depending on the string deformation at the moment it is hit by the hammer. Voice management also offers a lot of flexibility for less physically-realistic features such as dynamic changes of the model.

julien's Website

Reply by skip

  • Avatar
  • skip
  • Member
  • Offline
  • From: Sydney, Australia
  • Registered: 21-12-2007
  • Posts: 1,059

Re: Redoing string re-excitation to use only 1 polyphony

Julien: If the use of more than one "voice" results in Pianoteq producing something akin to "more strings", then it may not be as authentic as it could be. I think that's all we're saying.  Yes, overlapping voices, even in a ROMpler, produces a result that sounds quite realistic (to my ears) - the overlapping does indeed create interesting phasing effects that sound realistic. However,  if Pianoteq were able to model it by actually taking into account the state (shape, velocity etc) of the strings at the time of re-strike, the result might be even better. (I don't know).

Now, if you're saying that Pianoteq is doing something more advanced than simply overlapping multiple tone generators, and we should not make assumptions from the voice count, then that's fine.

One thing I have noticed that Pianoteq does is that it seems to dampen the sound in what seems a realistic way. For example, with the sustain pedal down, if you play a forte note, and then quickly play the same note pp, you can hear the sound level drop markedly when that pp note is played.  A simple ROMpler doesn't do this.

Greg.

Reply by julien

  • Avatar
  • julien
  • Admin
  • Offline
  • Registered: 19-04-2007
  • Posts: 1,026

Re: Redoing string re-excitation to use only 1 polyphony

skip wrote:

if Pianoteq were able to model it by actually taking into account the state (shape, velocity etc) of the strings at the time of re-strike, the result might be even better.

This is what I meant, Pianoteq does take into account the state of the string at the time of re-strike, it is not a simple overlap

julien's Website

Reply by skip

  • Avatar
  • skip
  • Member
  • Offline
  • From: Sydney, Australia
  • Registered: 21-12-2007
  • Posts: 1,059

Re: Redoing string re-excitation to use only 1 polyphony

Excellent!

Regards,
Doubting Thomas.

Reply by Beto-Music

  • Avatar
  • Beto-Music
  • Member
  • Offline
  • Registered: 02-05-2007
  • Posts: 2,786

Re: Redoing string re-excitation to use only 1 polyphony

There was a mp3 demo on pianoteq site demonstrating how fast repetitions produced diferente sounds than a simple repetition of the same initial tone vibration string sound.
Like in a real piano.

Last edited by Beto-Music (26-11-2013 15:27)

Reply by Philippe Guillaume

  • Avatar
  • Philippe Guillaume
  • Admin
  • Offline
  • Registered: 25-04-2007
  • Posts: 5,405

Re: Redoing string re-excitation to use only 1 polyphony

Beto-Music wrote:

There was a mp3 demo on pianoteq site demonstrating how fast repetitions produced diferente sounds than a simple repetition of the same initial tone vibration string sound.
Like in a real piano.

You're right , it's still there (4th audio example):
https://www.pianoteq.com/listen_features

Philippe Guillaume's Website

Reply by Beto-Music

  • Avatar
  • Beto-Music
  • Member
  • Offline
  • Registered: 02-05-2007
  • Posts: 2,786

Re: Redoing string re-excitation to use only 1 polyphony

Philippe Guillaume wrote:
Beto-Music wrote:

There was a mp3 demo on pianoteq site demonstrating how fast repetitions produced diferente sounds than a simple repetition of the same initial tone vibration string sound.
Like in a real piano.

You're right , it's still there (4th audio example):
https://www.pianoteq.com/listen_features


And this feature exist since the first beta version of pianoteq.

:-)

Last edited by Beto-Music (27-11-2013 01:08)