[Mp4-tech] Re: [Audio] CRC check in MPEG-4 ADTS stream

[Ralph Sperschneider]

MPEGIF List Admins wrote:
> Forwarding non-member reply that was auto-discarded.
>
> Thanks for the replay, and please remember to subscribe with your sending
> address (or to send from your subscribed account).
>
> the admins.
>
>
> -----Original Message-----
> From: Alexander Groeschel
> [mailto:Alexander.Groeschel codingtechnologies.com] 
> Sent: Monday, 17 April 2024 23:06
> To: klaus.raffold gmail.com
> Cc: mp4-tech lists.mpegif.org
> Subject: Re: [Mp4-tech] [Audio] CRC check in MPEG-4 ADTS stream
>
> Hi Klaus,
>
> mp4-tech-bounces lists.mpegif.org wrote on 04/17/2006 03:35:02 PM:
>
>   
>> Hi Experts,
>>
>> This is about implementing CRC check in MPEG-4 ADTS bitstream. 
>>
>> The ISO/IEC 13818-7 (MPEG-2 AAC) standard, where the ISO/IEC 14496-3, 
>>     
> subpart-1(MPEG-4 Audio) cross refers to for details on MPEG-4 ADTS, states 
> that ... 
>   
>> "
>> ...
>> If the length of a CPE is shorter than 192 bits, zero data are appended 
>>     
> to achieve the length of 192 bits. Furthermore, if the first ICs of the 
> CPE ends at the Nth bit (N<192), the first(192-N) bits
>   
>> of the second ICS are protected twice, each time in order of their 
>>     
> appearence. 
>   
>> ...
>> "
>>
>> What i understand from this is that, i have to parse the received 
>>     
> bitstream to find out which of the bits in the stream are CRC protected 
> and in what way. This means, that I have to start 
>   
>> interpreting the bits in the stream even before I get to know whether 
>>     
> they are corrupted or not. 
>
> Yes, unfortunately that's the case.
>  
>   
>> My parsing can go completely heywire if i start interpreting corrupted 
>>     
> bits. 
> Right.
>
>   
>> In that case what error protection does CRC offer in MPEG-4 ADTS stream 
>>     
> ?? 
> If the frame is corrupted, you still (with a certain security) will know 
> at the end that it's corrupted. But you're right, you waste time on 
> useless parsing attempts.
>
>   
>> OR is my understanding incomplete, or am I 
>> missing something ?? 
>>     
>
> There was an attempt inside MPEG to change this definition, which didn't 
> get approved - mainly because at that time chips already existed that 
> implemented this weired CRC calculation. Instead, the clarificaton cited 
> by you was added to the text.
>
>   
>> Thanks and regards,
>> klaus
>>     
>
> Hope this helps,
> regards,
> alex.
>
>   

Thanks to Alex for his response. One could state the facts also somewhat 
differently:
The intention of the CRC error detection is not to save the effort of 
decoding the audio frame beforehand in case of errors, but to prevent 
the output of corrupted audio data afterwards, particularly in cases 
where the error could not be detected by the parser. A prominent example 
for this would be a bit error in the MSB of the global gain.
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/