Topic by sigasa

  • Avatar
  • sigasa
  • Member
  • Offline
  • From: UK
  • Registered: 09-01-2008
  • Posts: 2,014

Topic: wave comparison analysis 'PianoTeq vs. Acoustic'

After visiting 'Pianobello' on the nweb last night, I started to think about the PianoTeq sound vs. the sound from a real piano. I have just sent a midi file to Pianobello for processing through a Steinway D in a studio. I will then compare the result with the C3 variations.

This gave me an idea (probably one that Modartt had years ago!!!). Wave comparison analysis for the purpose of finding the differences between PianoTeq's sound, and a real acoustic pianos output. I don't know if Modartt recorded all the original pianos that their models are based on. If they did, that would be brilliant. I'm thinking, is it possible yet to computer analyse two wav. files (PianoTeq/Real) and find the exact difference of the recording of the real Acoustic?

Then, once the differences have been analysed, then a way of implementing a 'model adjustment' by superimposing the differences onto the existing model, probably as a matter of programming in the ajustments to the mathmatical model(s)? Ans theoretically there you have it! Sound simple doesn't it? I bet it isn't!!!

Now I have this sneaky feeling that the Modartt team will be grinning from ear to ear at my naivity! They probably do wave analysis all the time anyway!?

Well that was my four penneth for 0131 in the morning. Going to sleep now! Bye!!!

Kind regards,

Chris

Reply by ReBased

  • Avatar
  • ReBased
  • Member
  • Offline
  • From: UK
  • Registered: 27-02-2009
  • Posts: 209

Re: wave comparison analysis 'PianoTeq vs. Acoustic'

Yes, that's what Modartt did to build their model.  The problem is you can't simply extract the differences, and apply them.  If you just wanted that, you could just make a sample-based instrument like everyone else.

Instead Pianoteq has to figure out how to model each physical component of the instrument.  The hammer & strings are the obvious and relatively simple part, but the body of the instrument, the sound board etc. also have to mathematically modelled. From what has been said, these can be described fairly accurately mathematically - but it is way too complex to do it in real-time on an average computer.  So Pianoteq uses many simplifications to get close to the better/slower ideal model while still making it playable.  That's the main reason why it still doesn't sound as rich as the real thing.

Of course simplifications can be improved, and code can be optimized or made faster with new techniques, so I'm sure PT4 will get closer to the real thing (I can't wait for PT4).

Last edited by ReBased (07-07-2011 11:30)

Reply by Beto-Music

  • Avatar
  • Beto-Music
  • Member
  • Offline
  • Registered: 02-05-2007
  • Posts: 2,783

Re: wave comparison analysis 'PianoTeq vs. Acoustic'

Interesting how pianoteq recreates better the sound of pianoforte, like Walter piano.

It make me guess what element in the modern Grand piano make it more comples to replicate by modelling.

Maybe the circular holes in the harp...