[FFmpeg-devel] [PATCH] lavf/ogg: Support for end trimming Opus

[Vignesh Venkatasubramanian]

On Mon, Sep 16, 2013 at 12:53 PM, Michael Niedermayer <michaelni at gmx.at> wrote:
> On Mon, Sep 16, 2013 at 10:08:57AM -0700, Vignesh Venkatasubramanian wrote:
>> On Sat, Sep 14, 2013 at 11:03 AM, Michael Niedermayer <michaelni at gmx.at> wrote:
>> > On Thu, Sep 12, 2013 at 10:24:18AM -0700, Vignesh Venkatasubramanian wrote:
>> >> Adding support for end trimming Opus embedded in Ogg container.
>> >>
>> >> Signed-Off By: Vignesh Venkatasubramanian <vigneshv at google.com>
>> >
>> > do you have some sample/testcase to test this patch ?
>> >
>>
>> i did the following tests for this patch:
>>
>> 1) encode a raw pcm file to opus and then decode it back to raw. the
>> file size of the
>> raw files should match exactly. without the patch, the decoded file
>> will be bigger than
>> the original file because of padding. i did this for 10 odd files.
>>
>> 2) Xiph has this file for checking correctness of Opus decoders:
>> https://people.xiph.org/~greg/opus_testvectors/correctness_trimming_nobeeps.opus
>> I played back this file using ffplay and it works as expected after
>> applying this patch.
>> Not before.
>
> thanks
>
> patch applied
>
>
>>
>> Do you want me to include fate tests for this? If so, could you please
>> provide some
>> guidance on how to do it? I am unsure how to implement a fate test for
>> this. Please
>> let me know.
>
> a fate test would be interresting, but iam not sure it is worth the
> effort
> it could be a demuxer only test
> (not sure how usefull that would be though)

yes, i don't think a demuxer only test will be useful here.

> or one that uses libopus, but that would be tricky if libopus
> doesnt have binary identical output accross platforms and libopus
> versions ...
>

from my understanding, libopus does not have binary identical output across
versions/platforms. may be someone who better knows about this can comment.

> [...]
> --
> Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
>
> I do not agree with what you have to say, but I'll defend to the death your
> right to say it. -- Voltaire
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>