[CELT-dev] STATIC_MODES, fft and prob members

Chen-Po Sun chenpo at fmod.org
Sun Jul 26 19:53:52 PDT 2009

Thanks for the quick responses Jean-Marc.

Looks like I'll be adding prob and fft to the static modes, just wanted 
a confirmation it wouldn't bust anything up. I'll look into sending some 
patches for the static probability stuff as well once I clean up the 
mess I've been making of the code.

I'm not sure why FFT isn't included even if the code is likely to be 
changed, the new code will still be generating the values to go into 
static_modes just like the old code would anyway.

Chen-Po Sun
Firelight Technologies

Jean-Marc Valin wrote:
> Chen-Po Sun a écrit :
>> I have noticed that even with STATIC_MODES turned on, celt_mode_create 
>> still allocs and generates the fft and prob members (via 
>> pitch_state_alloc and quant_prob_alloc respectively). I also notice that 
>> in dump_modes.c, these two members are just set to 0 explicitly. Why 
>> aren't these two members part of the static mode?
>> Would anything bad happen if I make them part of the static mode? I 
>> assume not because CELTMode is read-only right?
> The probability stuff is mainly because I've been too lazy so far (patch
> welcome!). OTOH, The reason why the FFT isn't included is because the
> FFT is the part of the code that's the most likely to be changed. It's
> very common to have a highly-optimised FFT that one can use instead of
> the generic one provided in CELT.
> Cheers,
> 	Jean-Marc

More information about the celt-dev mailing list