[sdiy] Prophet VS a phase accumulator design

[tu at alphalink.com.au]

> > *Beginning with the EMAX II and the Emulator IIIx, we used our "G-chip"
> > *technology, which was a 7th or 8th order interpolator operating from
> > *large fractional value phase accumulators. These gave excellent (still
> > *largely unsurpassed) alias and image attenuation, with all the advantages
> > *of a syncronous design. The original G-chip was 32 channels; the G-chip
> > *II had 64 channels per chip, and two chips could share a memory to give
> > *128 channels.

The G-chip in the Emax II still has a slightly grainy sound, I think from aliasing and sideband 
distortion around -60dB. Not that this is a bad thing musically, it gives a certain edge and bite to the 
sounds that cleaner samplers don't have. The G-chip II, as used in the Emulator 4, is much cleaner 
sounding and doesn't have the same character. I would be interested to know what sort of 
transposition was used by the Akai samplers, such as the S1000 and later. The Akai sound was 
clean and bright while still being able to shift upward by large intervals without audible aliasing.

> I noticed the Kurzweil K2000 series has pretty good interpolation, I wonder
> how they compare. It was mentioned recently on the music-dsp list that 
> "every" Alesis synth used linear interpolation. I'd a little be surprised
> if that included the Fusion as well, but that's at least the Quadrasynth 
> line. 

The K2000 has an excellent interpolator for transposing samples downward, or upward to the 48k 
system sampling rate, but there is no decimator! The maximum playback rate supported is 96k, so 
you can shift a 48k sample up an octave, but with that you get considerable aliasing distortion. I 
tested this once using a 20kHz sinewave sampled at 48k and a scope. As I went up the scale the 
sinewave went to nyquist at 24k and then the frequency started descending again. The K2500 
extended the upward transposition to two octaves at 48k using the "skip sampling" mode. I believe 
this mode did some crude filtering to reduce the aliasing from skipping samples but was 
acknowledged in the manual as having reduced fidelity.