[FFmpeg-devel] AAC-Main (round 2)

[Robert Swain]

2008/11/21 Alex Converse <alex.converse at gmail.com>:
> On Fri, Nov 21, 2008 at 11:30 AM, Robert Swain <robert.swain at gmail.com> wrote:
>> 2008/11/21 Alex Converse <alex.converse at gmail.com>:
>>> On Fri, Nov 21, 2008 at 10:51 AM, Robert Swain <robert.swain at gmail.com> wrote:
>>>> 2008/11/19 Alex Converse <alex.converse at gmail.com>:
>>>>> I split these off into a separate patch based on Rob's suggestion.
>>>>
>>>>> +            } else if (ac->m4ac.object_type == AOT_AAC_LC) {
>>>>> +                av_log(ac->avccontext, AV_LOG_ERROR, "Prediction is not allowed in AAC-LC.\n");
>>>>> +                memset(ics, 0, sizeof(IndividualChannelStream));
>>>>> +                return -1;
>>>>
>>>> Is this purely for informative reasons or have you encountered some LC
>>>> streams with the predictor_present bit set? This case doesn't exist in
>>>> the specs I have though I know it should not be allowed as prediction
>>>> is not in LC.
>>>
>>> I put that there because if the predictor bit is set in an LC stream
>>> then there is no sane assumption I can think of about the data that
>>> directly follows it.
>>
>> Good point. Just looking for some files encoded with Main profile with
>> which I can test, though I expect you have tested it thoroughly.
>>
>
> Try the am##_## series from
> http://standards.iso.org/ittf/PubliclyAvailableStandards/ISO_IEC_14496-4_2004_Conformance_Testing/audio_conformance/mpeg4audio-conformance/compressedMp4/

Thanks, I wondered which were main profile.

/me grins as Quicktime breaks on these files but patched ffplay works nicely.

I will commit the first three patches in a bit after casting an eye
over the code again.

Regards,
Rob