[Mp4-tech] Re: AAC LC mannual decoding of raw_data_block()

[Ralph Sperschneider]

goutham vaidhya wrote:
>
>
> On 5/15/07, *Ralph Sperschneider* 
> <ralph.sperschneider iis.fraunhofer.de 
> <mailto:ralph.sperschneider iis.fraunhofer.de>> wrote:
>
>     goutham vaidhya wrote:
>     >
>     > hi ,
>     >  The reply was very helpfull
>     >
>     > But I still have doubt.I have attached a file containing my
>     > understanding of windows, window groups , scalefactor bands &
>     sections.
>     > Could u pls have a look at it.
>     >
>     > one more thing, I want to mannually decode some more frames &
>     verify
>     > that it is right.
>     > So If u have any ADTS format file & the corresponding huffman
>     decoded
>     > spectral values pls send it.
>     >
>     > Thank you:
>     > Goutham N Vaidhya
>     > --
>     > Have a NICE day
>
>     Dear Goutham N Vaidhya,
>
>     in your attachment, you refer to 13818-7:1999. This does not
>     exist. The
>     first edition of MPEG-2 AAC was issued 1997, the second edition was
>     issued 2003. Tables with the number format x.y exist only in the first
>     edition, so I assume that you refer to this.
>
>     Your figure 4 is definitely wrong. Figures 2 and 3 are fine, if they
>     state the status prior to the deinterleaving process. Figure 1 is
>     fine,
>     if it states the process after the deinterleaving process
>     (although all
>     windows should then have the same size, but that might be just a
>     drawing
>     issue).
>
>     In the case of short blocks, you always have to deal with eight
>     spectra,
>     each of which has 128 lines total. This is independent of any
>     grouping.
>     So if, as in your example, max_sfb=11 (sampling frequency =
>     48kHz), then
>     in each spectrum the lines of the scale factor bands 12, 13 and 14
>     ( i.e.
>     the lines 80 till 127) need to be set to zero. The grouping only
>     reduces
>     the number of scale factors to be transmitted, implying that the same
>     set of scale factors has to be applied several times. At the same
>     time,
>     grouping leads to  sfb-wise interleaving of the codewords within
>     each group.
>
>     For further details I would like to refer to ISO/IEC 13818-7:2005
>     subclause 8.3.5 "Order of spectral coefficients in spectral_data" and
>     particularly Figure 6 "Spectral order of scalefactor bands in case of
>     EIGHT_SHORT_SEQUENCE" (in ISO/IEC 13818-7:1997 it is Figure 8.3).
>
>     ADTS sequences can be found here:
>
>     ftp://mpaudcon:adif2mp4@ftp.iis.fraunhofer.de/mpeg2aac-conformance/compressedAdts
>
.
>
>
>     The reference software can be found here:
>
>     http://standards.iso.org/ittf/PubliclyAvailableStandards/c039486_ISO_IEC_13818-5_2005_Reference_Software.zip
>
>     It should not that difficult to patch the AAC decoder so that it
>     outputs
>     the decoded spectral values. At the same time, the software should
>     help
>     you to figure out the correct interpretation of the AAC payload
>     and show
>     you the proper decoding steps.
>
>     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
>     <mailto:ralph.sperschneider iis.fraunhofer.de>
>     D 91058 Erlangen                | http://www.iis.fraunhofer.de/amm/
>
>
>  
> thanks for clearing my doubt. Now the concept is very clear.
> The link u have sent me is asking for username & password.

Sorry, the link should be:
ftp://mpaudconf:adif2mp4@ftp.iis.fraunhofer.de/mpeg2aac-conformance/compressedAdts 
<ftp://mpaudcon:adif2mp4@ftp.iis.fraunhofer.de/mpeg2aac-conformance/compressedAdts>
(username and pwd are already in the URL)
Best regards,
Ralph