[M4IF Technotes] Re: AAC Conformance

Ben Waggoner ben interframemedia.com
Wed Jun 11 11:01:14 EDT 2003


Ralph,
    Do I read this as saying that some AAC implementations could
meaningfully decode a file to PCM at more than 16-bits, ala Dolby Digital?
Ben Waggoner <http://www.benwaggoner.com>
Compressed Video Consulting, Training, and Encoding
My Book:            <http://www.benwaggoner.com/books.htm>
Cleaner e-book:     <http://www.cmpbooks.com/cleaner>
Compression Classes at Stanford  June 30-July 4 and Aug 11-15
<http://www.digitalmediaacademy.org/compression.html>
Compression Class in New York City, July 13
<http://www.manhattanvideoworkshop.com/courses.html>
on 6/11/03 6:20, Ralph Sperschneider at sps   iis.fhg.de wrote:
> for AAC, there is no such thing than "Fixed-point accuracy". However, there is
> an accuracy measure k for the two conformance tests measurements (RMS and LSB
> test). On default, k=16. If you are designing a decoder which is conform to a
> higher or to a lower level, you might (higher) / must (lower) specify this
> level 
> in your decoder specification.
> 
> This variableness has been introduced lately for both, MPEG-2 and MPEG-4 AAC.
> You will find the details in the appropriate conformance corrigenda:
> 
> * ISO/IEC 13818-4:1998/Amd.1:1999/Cor.1
>    (N5029, not yet published)
> 
> * ISO/IEC 14496-4:2000/Cor.2
>    (N5042, not yet published)
> 
> There, you will find the following text:
> 
> RMS/LSB Measurement
> To fulfill the ³RMS/LSB Measurement² test at an accuracy level of ³K bit², an
> ISO/IEC 14496-3 decoder shall provide an output waveform such that the RMS
> level 
> of the difference signal between the output of the decoder under test and the
> supplied reference output is less than 2^-(K-1)/sqrt(12).  In addition, the
> difference signal shall have a maximum absolute value of at most 2^-(K-2)
> relative to full-scale. The ³RMS/LSB Measurement² test shall be carried out
> for 
> an accuracy level of K=16 bit unless a different accuracy level is explicitly
> stated.



More information about the Mp4-tech mailing list