[Mp4-tech] Re: Doubt regarding MPEG4AAC

[Ralph Sperschneider]

lisa wrote:
> Hi Ralph,
>           I did not check the reference decoder software of ISO. I
> checked with faad2-0.
> 
> As I earlier mentioned for the files am02_32.mp4 to am02_96.mp4 there 
> are  small glitches that are coming on the right channel(in the faad code).
> 
> Can u please tell where the things CAN go wrong in that code?
> In the sense that which part of the decoder is being tested in that  
> particular set of test vectors?
> In which module the problems can possibly occur and for what possible  
> reason?
> What are the things that affect the decoding of the right channel?(in 
> the  sense that the left channel decodes properly)
> 
> Thanks in advance
> With regards
> Lisa.
> 
> 
> 
> On Tue, 07 Sep 2023 11:48:01 +0200, Ralph Sperschneider  
> <ralph.sperschneider iis.fraunhofer.de> wrote:
> 
>> lisa wrote:
>>
>>> Hi all,
>>>      I have doubt on the MPEG4AAC  main profile (audio) While 
>>> decoding  the  test vectors
>>>         am02_32.mp4  to  am02_96.mp4
>>>     There are some small glitches coming on the right channel. I 
>>> guess  thats  why the RMS test     fails.
>>>     Can anyone help on that problem? Remaining all test vectors are  
>>> passing.
>>>  thanx in advance
>>> lisa
>>
>>
>>
>> Hi Lisa,
>>
>> the current MPEG-4 Audio reference software decoder has no problems 
>> with  these sequences. Which decoder are you using?
>>
>> Best regards,
>>
>> Ralph
>>
> 
> 
> 

Hi Lisa,
I am not familiar with faad. I even cannot see something special which applies 
only to these sequences. The sequence was intended to test as many different 
Huffman codewords from the spectrum Huffman codebooks as possible (that is 
called "arithmetic torture" in ISO/IEC 14496-4). Maybe some seldom used Huffman 
codes (eventually with some escape sequences) are not correctly decoded by the 
faad ... just a guess.
Best regards,
Ralph
-- 
Dipl.-Ing. Ralph Sperschneider  | Phone: +49 9131 776 344
Fraunhofer IIS                  | Fax:   +49 9131 776 398
Am Wolfsmantel 33               | mailto:ralph.sperschneider iis.fraunhofer.de
D 91058 Erlangen                | http://www.iis.fraunhofer.de/amm/