It would be very interesting to create a program that analyzes sound filesin some other way than FFT.
I have all the FFT tools I could ask for (Kyma, Csound, Lemur, audio sculpt),but I'm still searching for a more pure way of creating additive synthesis.FFT has a way of coloring the sound (not necessarily in a bad way, but still..).
The Synclavier II uses 96 operator additive synthesis, where you load spectralsets from floppy's. Quasi sampling almost. However the sound is fantastic!Much more pure than any FFT software of today. Now Synclav might have usedFFT to analyze the samples, but somehow they managed to keep the time domainfree from wavering and flutter. A software that could generate tables likethat would be amazing!
How about a more un orthodox way of creating additive spectra's. What ifyou would use a vocoder approach and record the rms for each band.
This requires a lot of filters, but it wouldn't have to be real time. Infact it's almost to prefer if it wasn't in real time, because then you coulduse better sounding filters.
Tobias
tontaub wrote:
Hi Paul,
--- In motm@yahoogroups.com, Paul Haneberg <phaneber@o...> wrote:
> I am thinking of writing my own FFT analysis program for analyzing
> wave files. If I do I will post a link to the group. I intend to
> share it for free.
if you do so, how are you going to deal with for instance f-modulation
of the fundamental? I'm asking because IIRC all those programs cut the
sampled signal into chunks of 2^n samples regardless of its frequency.
Due to the applied windowing techniques (Hamming, Hanning, Bartlett
etc.) the results are mathematically (more or less) correct but the
displayed spectras are IMHO not very useful in a musical sense.
Granted, many things are detectable by the ear but in my case I'm
still seeking automated help to get even closer to the desired results.
;-) Michael.
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.