[sdiy] FW: Kawai K3 - Bandlimited Waveforms

[Tom Wiltshire]

On 17 Jan 2010, at 22:34, karl dalen wrote:

> Its interesting to note that the K3 does a poor job in this
> sample table per octave switching, one can very easily hear
> when a table are swapped in and out.

They didn't use any interpolation between adjacent wavetables then.
As we've already discussed, they did interpolate between wavetable  
samples, but it looks like that was it.

> Btw, should phase accumulator NCO's be wieved as
> variable sample rate or fixed rate NCO's?
>
> I was thinking of the aliasing on an NCO using low rate
> versus one that uses a high rate (1Mhz or more).

Well, this is where the confusion arises.

I'd say (and this is an opinion not a statement of fact) that if it's  
got a "high" sample rate and "small" tables, then it can be viewed as  
a variable sample rate system. At least, it approximates one. This is  
the PPG scheme. In a system like this, each individual wavetable  
sample is typically repeated many times, which is roughly equivalent  
to clocking the samples out at a much slower, variable rate. In  
practice, the line blurs, since at the top of the PPG's range, it  
starts missing samples out because the sample rate isn't high enough  
(only 250KHz).

Obviously the reality is that both systems have a fixed rate, and  
it's purely a way of viewing it.

T.