Jim Menard <jimm at io.com> wrote:
> How about doing what the Roland D-50 introduced commercially (I've
> forgotten the correct name): play a sample at the beginning of a note then
> use a simpler, repetitive waveform for the rest of the note. It's the
> complex wave in the first 100 to 300 ms that the ear uses to identify the
> instrument.

LoopedSampledSound supports precisely this functionality. The hard part
is that a one-cycle loop sounds static and dead, since it has none of the
little fluctuations of pitch, timbre, and loudness that characterize a real
musical instrument. Some of the "life" can be put back in (we hope) via
pitch and loudness envelopes, perhaps with some randomness, and
some of the timbral dynamics might be added via a variable lowpass filter
or by using FM wave distortion techniques. There are lots of things to try,
which makes it fun! Of course, the major synth vendors have been working
on this very problem for at least a decade, but they tend to keep the details
of their best ideas to themselves for some reason...  :->

I don't expect to find a trick that the synth manufactures have missed, but
I'll be happy if we re-discover some known techniques that give us good quality
sounds in Squeak. And you never know where the ability to explore sound
synthesis in a dynamic environment like Squeak may lead...

	-- John