[FFmpeg-devel] [PATCH] QCELP decoder

[Kenan Gillet]

On Nov 14, 2008, at 9:37 AM, Vitor Sessak wrote:

> Michael Niedermayer wrote:
>> On Thu, Nov 13, 2008 at 11:46:40AM -0800, Kenan Gillet wrote:
>>> Thanks for your reviews,
>>>
>>> here is round 10 of the QCELP decoder patch:
>>> - it simplifies and optimizes qcelp_lspf2lpc and lps2poly
>>> - it removes bandwith_expansion_coeff from QCELPContext, and  
>>> replaces it
>>>  by a systematic recalculation as the benchmarks show it is faster
>>> - simplifies apply_gain_control,
>>> - adds FIXME comment an av_log_missing_features in  
>>> apply_gain_control
>>>  if the vector to control gain of is a zero vector.
>>> - various cosmetics
>>>
>>> Kenan
>>>
>>
>> [...]
>>
>>> @@ -461,6 +517,15 @@
>>> #define QCELP_RATE_HALF_CODEBOOK_RATIO 0.5
>>>
>>> /**
>>> + * sqrt(1.887) is the maximum of the pseudorandom
>>> + * white sequence used to generate the scaled codebook
>>> + * vector for framerate 1/4.
>>> + *
>>> + * TIA/EIA/IS-733 2.4.8.1.2
>>> + */
>>> +#define QCELP_SQRT1887 1.373681186
>>> +
>>> +/**
>>>  * table for impulse response of BPF used to filter
>>>  * the white excitation for framerate 1/4 synthesis
>>>  *
>>
>> ok
>
> ok'ed parts applied.

thanks