[Mp4-tech] Re: [audio] best coder for 64kbit/s

[Ralph Sperschneider]

Noor Othman wrote:
> Dear Ralph Sperschneider,
> 
>    Thank you for your reply. At the moment, I have yet to understand 
> what is the different between 1 channel or multi-channel.

In the uncompressed (pcm) domain, each channel contributes to the overall 
bitrate equivalently. Hence, a stereo wav-File of a certain duration has twice 
the size of a mono wav-File of the same duration, assuming that the same bitrate 
and the same sample-coding has been used.
In a general point of view, this is similar in the compressed domain. However, 
the audio coder might be able to reduce the overall bitrate using joint coding 
tools (the most common one in the case of AAC is M/S coding).
> If you don't 
> mind, can I ask you about stereo sample file, which has 2 channels, and 
> how do I pass this into AAC coder? Or should I combined both, to be 
> mono, and then just pass it through 1 or single channel?

The encoder has to be informed about the number of channels available in the 
input file and the number of channels it shall encode.
>    What does this multi-channel do? Splitting the info? Can a mono audio 
> input file, being passed through a multi-channel encoder? I am in a 
> great confusion... Thank you so much if you can help.

A multi-channel encoder is usually able to encode up to n channels. So yes, it 
can also encode mone signals.
Ralph
> 
> Noor
> 
> 
> Ralph Sperschneider wrote:
> 
>> Dear Noor,
>>
>> Noor Othman wrote:
>>
>>> Hi,
>>>
>>>    Can anyone help me to understand, what is the best coder at 
>>> 64kbit/s and above?
>>
>>
>>
>> Do you mean 64kbit/s/channel?
>>
>>
>>> And did I get it right that TwinVQ is only operates at bit rates 
>>> between 6-16 kbit/s?
>>
>>
>>
>> You are right with respect to what is permitted in the MPEG Audio 
>> standard.
>>
>>
>>> But I have tried for higher bitrate it gives better output.
>>
>>
>>
>> TwinVQ generally operates also for higher bitrats and as you found the 
>> quality still increases with the number of bits spend. However, above 
>> 16 kbit/s it is recommended to use AAC since it provides better 
>> quality at those bitrates than TwinVQ. TwinVQ uses a vector quantizer 
>> for the redundancy reduction, but AAC uses Huffman coding. Huffman 
>> coding is lossless, while the vector quantizer is not.
>>
>> 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/