[Ffmpeg-devel] [DRAFT] Add fact chunk to non-PCM wav

Michel Bardiaux mbardiaux
Wed Feb 7 17:55:54 CET 2007

Michael Niedermayer wrote:
> Hi
> On Thu, Nov 09, 2006 at 01:53:08PM +0100, Michel Bardiaux wrote:
>> Michael Niedermayer wrote:
>>> Hi
>>> On Thu, Nov 09, 2006 at 11:36:13AM +0100, Michel Bardiaux wrote:
>>>> Michael has reported that
>>>>>> interresting, accoridng to microsofts excelent and unambigous 
>>>>>> documentation
>>>>>> (not kidding ...)
>>>>>>    Fact Chunk
>>>>>> This chunk is required for all WAVE formats other than WAVE_FORMAT_PCM. 
>>>>>> It stores file
>>>>>> dependent information about the contents of the WAVE data. It currently 
>>>>>> specifies the time length of the
>>>>>> data in samples.
>>>>>> so this must not be under CODEC_ID_MSGSM, also it must be a seperate 
>>>>>> patch
>>>>>> as its not CODEC_ID_MSGSM specific
>>>> I took this literally, hence CODEC_ID_PCM_ALAW and CODEC_ID_PCM_MULAW 
>>>> will get a fact chunk too.
>>> :)
>>> now just take the largest pts minus the smallest pts of any packet stored
>>> and convert that by using AVStream.time_base and AVCodecContext.sample_rate
>>> to the number of samples and store that in the fact chunk
>> Right, the patch does not make sense without the final update of the 
>> chunk, and with a *general* formula, not the one I had specialised for 
>> MSGSM. I'm rather rusty on the handling of timestamps, so this might 
>> take some time 
> in the write packet function
> assert(avpacket->pts != AV_NOPTS_VALUE);
> context->maxpts= FFMAX(context->maxpts, avpacket->pts);
> context->minpts= FFMIN(context->minpts, avpacket->pts);
> and in write_trailer
> number_of_sample= av_rescale(context->maxpts - context->minpts, avctx->sample_rate * (int64_t)avstream->time_base.num, avstream->time_base.den);
> untested of course but it should work approximately that way ...

Thanks, it works, with a little correction: the duration of the last 
packet must be added to the difference in pts. I assumed the 
pkt->duration is in the same unit as the timestamps.

>>> then run the regression tests and send a patch which updates the checksums
>> Not clear how to proceed here, do you mean a subsequent patch, or in the 
>> same?

I havent forgotten to make test, and of course the changes breaks 
regression since it adds 12 bytes to every non-PCM wav. I will post the 
final version with new checksums when the rest is approved.

Michel Bardiaux
R&D Director
T +32 [0] 2 790 29 41
F +32 [0] 2 790 29 02
E mailto:mbardiaux at mediaxim.be

Mediaxim NV/SA
Vorstlaan 191 Boulevard du Souverain
Brussel 1160 Bruxelles

More information about the ffmpeg-devel mailing list