On 11-Jun-99 Gabriel Bouvigne wrote:
> "As the audio signal is real, a FFT of the half order is sufficient to
> derive the spectrum, provided that a few linear combinations be used at both
> sides of the FFT. In the case of a stereo signal, one can also implement a
> single full order complex FFT to compute the spectrum. Further details about
> those algorithms may be found in [3].
At the moment LAME uses 2 half order (real only) FFTs to compute the two
channels.
I'm not sure it'd be worth trying to use 1 full order to compute both at the
same time.
but i've been known to be wrong before.. :)
later
mike
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
