On Mon, 28 Apr 2014 10:05:42 +0200, Hendrik Leppkes <[email protected]> wrote:
> On Mon, Apr 28, 2014 at 7:34 AM, Anton Khirnov <[email protected]> wrote:
> >
> > On Sun, 27 Apr 2014 23:06:07 +0200, Hendrik Leppkes <[email protected]> 
> > wrote:
> >> On Sun, Apr 27, 2014 at 6:47 PM, Anton Khirnov <[email protected]> wrote:
> >> > ---
> >> >  libavformat/matroska.c |    1 +
> >> >  1 file changed, 1 insertion(+)
> >> >
> >> > diff --git a/libavformat/matroska.c b/libavformat/matroska.c
> >> > index 410e2f4..9628abc 100644
> >> > --- a/libavformat/matroska.c
> >> > +++ b/libavformat/matroska.c
> >> > @@ -32,6 +32,7 @@ const CodecTags ff_mkv_codec_tags[]={
> >> >      {"A_MPEG/L2"        , AV_CODEC_ID_MP2},
> >> >      {"A_MPEG/L1"        , AV_CODEC_ID_MP2},
> >> >      {"A_MPEG/L3"        , AV_CODEC_ID_MP3},
> >> > +    {"A_OPUS"           , AV_CODEC_ID_OPUS},
> >> >      {"A_PCM/FLOAT/IEEE" , AV_CODEC_ID_PCM_F32LE},
> >> >      {"A_PCM/FLOAT/IEEE" , AV_CODEC_ID_PCM_F64LE},
> >> >      {"A_PCM/INT/BIG"    , AV_CODEC_ID_PCM_S16BE},
> >> > --
> >>
> >>
> >> This should only be added if you also write the Opus specific
> >> information for SeekPreRoll and CodecDelay when muxing (or muxing
> >> disabled until that is so), otherwise you produce non-standard files.
> >> You could port these muxing things from FFmpeg, fwiw.
> >>
> >
> > I don't see why would those elements (or the problem they solve) be in any 
> > way
> > Opus specific.
> 
> "Specific" was probably not the exact right term, but they were
> introduced for proper Opus support, and without at least CodecDelay,
> you won't be able to produce proper timestamps for Opus audio in MKV.
> SeekPreRoll is IMHO not as crucial.

My point was that this exact problem exists for any other MDCT-based codec.
So not writing CodecDelay for Opus is exactly as wrong as not writing it for
MP3, Vorbis, AAC, etc.

Anyway, I'll send a separate patch that writes the delay.

-- 
Anton Khirnov
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to