[FFmpeg-devel] [PATCH] check sampling index validity when parsing adts frame header

Alex Converse alex.converse
Tue Feb 10 18:58:38 CET 2009


On Tue, Feb 10, 2009 at 8:48 AM, Robert Swain <robert.swain at gmail.com> wrote:
> 2009/2/9 Robert Swain <robert.swain at gmail.com>:
>> 2009/2/9 Jai Menon <jmenon86 at gmail.com>:
>>> On Mon, Feb 9, 2009 at 1:51 PM, Baptiste Coudurier
>>> <baptiste.coudurier at gmail.com> wrote:
>>>> Jai Menon wrote:
>>>>> On Mon, Feb 9, 2009 at 12:41 PM, Alex Converse <alex.converse at gmail.com> wrote:
>>>>>>> Index: libavcodec/aac.c
>>>>>>> ===================================================================
>>>>>>> --- libavcodec/aac.c  (revision 16998)
>>>>>>> +++ libavcodec/aac.c  (working copy)
>>>>>>> @@ -1529,6 +1529,11 @@
>>>>>>>          if (hdr_info.chan_config)
>>>>>>>              ac->m4ac.chan_config = hdr_info.chan_config;
>>>>>>>          ac->m4ac.sample_rate     = hdr_info.sample_rate;
>>>>>>> +
>>>>>>> +        if(hdr_info.sampling_index > 11) {
>>>>>> 12 is a valid sampling index. It corresponds to 7350 Hz.
>>>>>
>>>>> Shouldn't it be in the table then?
>>>>
>>>> Isn't it ? check mpeg4audio.c
>>>
>>> the table I referred to is swb_offset_1024 in aacdectab.h
>>
>> I'll have to check the spec for this. There are other tables which are
>> indexed by sampling_index ( swb_offset, ff_aac_num_swb, tns_max_bands,
>> ff_aac_pred_sfb_max as well as ff_mpeg4audio_sample_rates ). If these
>> other tables only support up to index 11 in the spec then that's what
>> we shall have to check. If they should be extended and for whatever
>> reason they were truncated, we'll extend them.
>
> From the spec, it seems that a sample rate of 7350Hz (sampling_index
> == 12) is not supported in AAC. At least, none of the above mentioned
> tables contain entries for 7350Hz, only for 8kHz and above. In which
> case, I consider the checks valid.
>

"Table 4.68 - Sampling frequency mapping" (1496-3:2005) says that for
f<9391 the tables for 8000 Hz should be used.

[...]

--Alex




More information about the ffmpeg-devel mailing list