Re: [fluid-dev] fluid~ crashes

[Tim Goetze]

[Josh Green]

>On Wed, 2004-03-03 at 02:56, Peter Hanappe wrote:
>> Garett Shulman wrote:
>> > While I'm at it... This is probably completely outside the paridigm of
>> > fluidsynth, so if you are offended by the suggestion please just
>> > disregard... Really, the only thing that is keeping fluidsynth from
>> > completely emulating subtractive synthesis is pwm. It would be usefull
>> > to have a generator that could take an offset from the start of a sample
>> > and set all output of that sample to 0 after that. Or something like
>> > that. But, perhaps such a beast does not belong in the main fluidsynth
>> > source. Perhaps a patch would be more appropriate.
>>
>> Do you think PWM is possible using the modulation envelope? You could
>> control the width by changing the decay and sustain phases in runtime.
>> The volume envelope can not drop to zero instantaniously but in a span
>> of 64 samples.
>>
>> Cheers,
>> P
>
>My impression of the abbreviation "PWM" is Pulse Width Modulation. Can
>anyone clue me into what this is and how it would be used with SoundFont
>synthesis? Just curious.

as you probably already guessed, it's usually applied to square waves
as used by analog synth gear and describes the relation of the width
of the upper lobe to the lower lobe of the square waveform.

there are a number of ways to apply it to sampled waveforms, with no
fixed semantics. they are all a bit problematic since usually the
sampled waveform loop consists of a lot of cycles, not just one,
which would be much easier to modify towards a square wave.

the best thing, i think, would be to add analog-style oscillators to
fluid to realize Garett's request. it's a complicated business though
(you have probably already noticed that sampled analog-style waveforms
can alias very badly).

vriendelijk,

tim