Subject: Re: [linux-audio-dev] Resampling
From: Juhana Sadeharju (kouhia_AT_nic.funet.fi)
Date: Wed Aug 01 2001 - 22:40:03 EEST
>From: Alexander Ehlert <alexander.ehlert_AT_uni-tuebingen.de>
>
>I just do an n-times oversampling DFT. You can configure blocksize
>and oversampling(overlap) factor. Then for the resampling process
>I just add/remove bands in the frequency domain and then do resynthesis.
>Thats it. In the oversampling and resynthesis code I use Hanning windows.
OK. I have been quite skeptical to any FFT analysis/synthesis because
the phase vocoder in Csound sounds very muddy even no change is done.
But if you're handling only the complex values got from FFT (not amplitude-
phase versions), then everything should be better.
I have planned to do the following test:
1. Generate a set of random frequency/amplitude/phase triples below
some frequency Fc.
2. Using sin() generators, one per above triples, output a float
signal to file with (i) source and (ii) target frequencies.
3. Resample the file (i), and compare the resampling against the
file (ii).
The test could be done separately for different frequency bands
so that inaccuraties in high frequencies don't mix with the results
got for low frequencies. Low frequencies could be tuned better.
>BTW: Yesterday I gave my flanger effect some polishing, you can now
>change parameters realtime and it has a nice feedback and gain control.
If you're more interested in effects, you could check Kurzweil 2500's
Algorithms manuals from
ftp://ftp.youngchang.com/pub/Kurzweil/Pro_Products/K2000-K2vx-K2500/
K2500/Documentation/KDFXALGS.ZIP
It at least gives the whole picture of what effects are needed in
a good synth system. There are also user's manuals and reference manuals
for Kurzweil synths.
Regards,
Juhana
This archive was generated by hypermail 2b28 : Wed Aug 01 2001 - 22:39:56 EEST