[sdiy] So called "LA SYNTHESIS" : Was: LA Synth Nuts...

[jhaible]

> Actually, that was one of the main things that really frustrated me about
> the D50. I really think that allowing the sampled waveforms to be filtered
> would have opened up a lot of additional possibilities for the instrument.
I
> never understood why they decided to eliminate the filter when using the
> samples.

I suspect they _had_ to make it that way. The "filter" is, more or less,
just
a triggered (truncated?) impulse response - triggered by the pulse wave.
If you wanted a saw wave "filtered", that PW-triggered impulse response
was multiplied afterwards with a sine wave. (Not the same as a filtered saw,
but similar enough.)
If we assume that the machine had the capability to *either* handle a stored
sample (per layer) *or* a filter impulse response at the same time, this
odd limitation suddenly seems to make sense.
Also, the real time multiplications were very restricted. If you have
a ring modulation in your patch, the multiplication must be missing
somewhere else. I don't have the algorithms in front of me (as printed
on the D-50), but if memory serves with RM you had only one DCF left
per *pair* of partials.

JH.