[music-dsp] Re: lossless pcm compression

[Joshua Scholar]

>
> I think you are confusing redundancy with irrelevance. Irrelevant
information (small imperceptible differences) are rejected in the
> process
> of lossy compression, but this is not allowed in lossless compression,
wherein you may only remove the redundancy.
>
> While there is a lot of correlation is quasiperiodic waveform, the entropy
is still high, due to the random component, as pointed by
> Joshua.
>

No I think he was pointing to the fact that you should be seeing the same
samples over and over again from your synthesizer.  However blind separation
of a signal into the optimum sum of repeating samples isn't a problem that
anyone has solved.