[sdiy] FPGA Synth Music

[cheater cheater]

On Mon, Mar 8, 2010 at 17:17, Scott Nordlund <gsn10 at hotmail.com> wrote:
>
>> I'm not sure about the Synclavier or the Con Brio, but one really interesting thing about the Synergy is the flexibility of its envelopes, both the number of stages and the way you can crossfade between envelope parameters both in velocity and in key note.
>
> I believe the K150's envelopes are similarly sophisticated.  The K150 patent is 4833963, but it's not really anything that you wouldn't expect.  It appears that since they don't really have to worry so much about aliasing (since it's just sine waves, no FM), they thought it was advantageous to go with a quite low sample rate (19.53 kHz).  Aside from that it's the expected pipeline/log sine thing, with the addition of a noise ROM.  They actually worked this additive engine into the first home piano Ensemble Grande models, since it was (briefly) cheaper than sample based stuff.
>
> I think the Synergy still has an advantage, though, in that it has independent pitch modulation per partial (if I'm not mistaken).
>
> Interesting the way things seemed to go in those days: a basically open ended synthesis architecture that relies on software to do all the important stuff, but with a ridiculous amount of analysis and data to make a single decent sound.

Did the Synergy have a way of taking PCM samples and analyzing them
down into FM operators with pitch envelopes? I am not familiar with
the synthesis engine.

>
>> On another note, I've always been curious to see the guts of the original Synclavier...
>
> This is a block diagram of an FM voice: http://www.500sound.com/SyncII/synthschem.jpg  Unlike all of the "smart" designs, it seems to use hardware multipliers and separate hardware for each voice.  There is some quality advantage to this, since it supports higher sample rates,

I guess that made sense when the hardware was just slow, but with
today's cpus we can really run that at humongous sample rates AND huge
polyphony.

> and each voice has its own multiplying DAC,

There's no reason why you can't do that with the pipeline approach. In
fact Yamaha DX synths had stereo outs; some rack had 8 separate
outputs, one for each multi. But it costs some extra silicon.

> but it's nasty 8 bit stuff anyway.

8 bits for one voice is not so bad. Bear in mind that many synths
nowadays have whole orchestral ensembles coming out in 16 bit, and
that's supposed to sound good. Those 8 bits of that one voice will
might never exceed 8 bits of information on the final medium.

>
> I've skimmed the Synclavier manuals as well.  The "additive" stuff is really not very exciting, you can only define a fixed single cycle waveform with 24 harmonics.  The interesting stuff (including resynthesis of sampled sounds) is done by software, where it crossfades between two of these waveforms, calculating and storing a new waveform to the "offline" voice when it's all the way faded out.  All of the good sounds seem to be made by stacking tons of voices together.
>
> ...Now what about the Technos Axcel stuff?  Or even Kawai K5?
>
> _________________________________________________________________
> Hotmail: Trusted email with powerful SPAM protection.
> http://clk.atdmt.com/GBL/go/201469227/direct/01/
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>