[Mp4-tech] Re: regarding TNS and Dependent Coupling

[Ralph Sperschneider]

goutham vaidhya wrote:
> hi ,
>   Is these things right or wrong pls tell me.
> 1) The input to the Dependent coupling block should it be interleaved 
> i,e, in coeff[g][win][sfb][bin]?
> 2) The input to the TNS block should be deinterleaved i,e,, 
> coeff[win][bin]
> In the function (given in standard 13818:1997)
>   --> couple_channel(source_spectrum[],dest_spectrum[],gain_list_index)
>        here the source_spectrum[] is what we get from 
> individual_channel_stream() from CCE right?
>        the dest_specrtum[] is what we get from the TNS block( assume 
> TNS block occurs)
>        What the spectrum we get from the TNS block is in the form of 
> coeff[win][bin]
>        But in the dependent coupling computation we access the 
> spectral coeffs in coeff[g][win][sfb][bin] right?
>     
>       i want to know if we can access the spectral values directly Or 
> is there some modifications that has to be applied to the function 
> couple_channel()
>     
>      awaiting earliest reply.Tnx
>  

Dear Goutham Vaidhya,
the issue you describe is an implementation issue. You might want to 
interleave and deinterleave the spectrum several times or you simply 
calculate the proper index depending on the kind of interleaving 
required in a particular signal processing block. The latter is done in 
the reference software. I already provided you with the appropriate link 
(http://standards.iso.org/ittf/PubliclyAvailableStandards/c039486_ISO_IEC_13818-5_2005_Reference_Software.zip).
I suggest that you have a closer look into this software (in 
aacDec/src/adifdec.c, line 740ff, the functions coupling() and tns() are 
called).
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/