Re: Le mat?riau Sonore, hidden place

Subject: Re: Le mat?riau Sonore, hidden place
From: Michael Gogins (
Date: Fri Sep 16 2005 - 09:45:46 EDT

Until recently, computers have not been fast enough to simulate the complexity of the grand piano.

This will change. A Linux cluster should now be capable of modelling all the strings, the soundboard, and some of the hammer striking sounds, and of course the recital hall -- not in real time of course, but in a usable time. Probably do several live notes per workstation in the cluster, or something like that.

To get a picture of what this might mean, check out the analogous software for the pipe organ, Aeolus (of course the pipe organ is much easier to model than the grand piano). This software presents a good model of an 18th century pipe organ. Like many accurate digital simulations, it is "hyper-real", it has all the features of the real thing but in a pristine, noiseless space, and with less "movement" within the sound.

One wouldn't mistake Aeolus for a real organ, at least not if one had recently HEARD a real organ, but one can probably use Aeolus to make a musically interesting rendering of Bach or Franck or whatever, which would have been impossible before. I would say it is much better than an electronic church organ.

It would in fact be challenging to render some Bach or Franck or Messien by taking complete control over dynamics, phrasing, registration, etc. in ways that would not be possible for a live player.

I think it would be possible to get the movement within the sound with some additional work, also, perhaps convolving with a changing impulse response.


-----Original Message-----
From: Richard Wentk <>
Sent: Sep 16, 2005 9:13 AM
Subject: Re: Le mat?riau Sonore, hidden place

At 01:35 16/09/2005, you wrote:

>well, it's not a contest right? and it depends on what
>you mean by 'broader' and 'pallet'. i think what
>ned's talking about is how complex, say the sound of a
>piano note is when you play a key and listen to the
>sound of it decay...the sound is incredibly complex,
>resonates with the other strings, different harmonics
>(or partials, or overtones, whatever) reinforce and/or
>cancel each other, and this all occurs over a period
>of time and is influenced by how well it's tuned, how
>hard you hit the key, the temperature in the room,

I listened to someone performing professionally on a top of the line
Bosendorfer recently, and it was *the* most amazing thing. The initial half
second or so of each chord was perfectly in tune (or as in tune as anything
can be in 12ET), and then all the partials seemed to decorrelate
simultaneously to produce a fantastic chorus and bloom.

I've listened to quite a few pianos over the decades, and it's the first
time I've ever heard a sound quite like that.

> to try to do that with a computer, well, we're
>not there yet. and i'm not sure i'd want to be..there
>are plenty of things a computer can do that a piano

But there's always the issue of quality versus quantity. Begin able to make
an endless variety of sounds becomes less interesting if few are musically
worthwhile. Or if they're less musically worthwhile and satisfying than
acoustic instruments.

There's a *huge* amount still to find out about the intricacies and nuances
of acoustic instruments, both as sound generators and as performance tools.
I think it's a shame to concentrate on virtual experiences so much when
acoustic experiences still remain richer and more interesting.

Bridging the gap is a good thing to aim for, but I think that's going to
need much better tools and more imaginative thinking about synthesis than
we seem to have available today.


This archive was generated by hypermail 2b27 : Sat Dec 22 2007 - 01:46:11 EST