[FFmpeg-devel] [PATCH]Support qclp in mov before stsd version 1

Paul B Mahol onemda at gmail.com
Thu Jun 6 12:55:40 CEST 2013


On 6/6/13, Paul B Mahol <onemda at gmail.com> wrote:
> On 6/6/13, Carl Eugen Hoyos <cehoyos at ag.or.at> wrote:
>> Paul B Mahol <onemda <at> gmail.com> writes:
>>
>>> On 6/6/13, Carl Eugen Hoyos <cehoyos <at> ag.or.at> wrote:
>>> > Paul B Mahol <onemda <at> gmail.com> writes:
>>> >
>>> >> > Attached patch fixes the following sample as reported by Compn:
>>> >
>>> >> You sure this does not break other qcelp in mov files?
>>> >
>>> > It does not break 0407090925b.3g2, not even if I
>>> > set incorrect values for bytes_per_frame.
>>
>> (Meaning: Both possible and impossible values.)
>>
>>> But does such file actually use variable frame size
>>
>> I haven't seen a variable frame size sample so far.
>
> Thing is: audio packets in this file contain more than one audio frame.
>
> So i wonder if qcelp, and maybe evrc should have CODEC_CAP_SUBFRAMES
> added or this kind of muxing is done when no variable rate frames are
> present.
>
> So your patch may just fix one case, and leave others still broken.
>
> I'm not against this patch to be applied, but its clearly not optimal
> solution.

Forgot to said: make sure before applying patch that no regression in decoding
is introduced for files that actually have each frame into separate
packet and thus
use variable frame size - packets are of different size.


More information about the ffmpeg-devel mailing list