J.P.Fitch@bath.ac.uk wrote:

> I have added to my experimemtal system a family of opcodes which are
> like oscili except that they use cubic interpolation.  If anyone withs
> to experimemt with these I can provide SGI, Linux or PC versions, or
> of course code.  I am not totally sure that the code is correct in all
> cases but it seems OK.  At present I have called this family oscil3
> for cubic.

Great! it is very intriguing.
 
> Question: should this code replace oscili (assuming it is correct)?

Absolutely no! It should be slower, and I think that this kind of interpolation will be
useful, above all, for wavetable-based synthesis (so I need it on an opcode such as
loscil. You can name it something as 'cuboscil').

> Wouks you prefer oscil3 to work with non-power-of-2-plus-1 tables?

Yes!
 
> Question: which other opcodes do you think should use cubic
> interpolation?  Adjusting an opcode is not trivial, and certainly is
> much slower.  

diskin!

I am willing to do some more opcodes, but I woudl prefer
> a priority order -- or promises that people with do teh adjustments
> AND SEND BACK THE CODE.

> 
> Question: is cubic needed?  What about using quadratic interpolation?

what about a 8-point interpolation (such as Sound Blaster Live! wavetable synth)?


> PS Happy new year

Happy 1999 you too!
-- 
Gabriel Maldonado