EXS 24 Logic Sampler Users Group

Sascha Franck wrote: 

>
>  I was a bit astonished about how much aliased some of the VSTi's
> bounces were, almost completely unusable when stripped down to single
> waveforms. 


Rene's waveforms for Pentagon 1 -- look at the wavetables in an editor -- might
make a good source. They are alias free and there is a cycle for each midi note
(128 of them). 

>
> Actually not even WaveLab's signal generator does the job that
> well, you loop one single waveform of a longer file and it's outta tune,
> then you loop another one and again it's outta tune, just in a different
> way - weird!


I would expect short wave samples to be out of tune due to sampling being based
discrete numbers of samples.  So a 1kHz wave is really only 44 samples per
cycle if you are sampling at 44.1kHz. If you loop 44 samples you will get a
periodic function that is 0.2% too short. A 6%frequency change  is a semitone
so you are likely to be something like 3 cent sharp on that example. 

Better tuning would result from more cycles looped say 10 or more at higher
frequencies -- pitch errors are smaller for bass notes of course since they
will contain more samples per cycle. 

Regards,
Murray