Reply by pkon...@gmail.com●April 19, 20052005-04-19
hi wang,
For simulation purpose three files are given for EFR speech codec. One is coder,
that compresses 160 samples to 244 bits. The intermediate module converts 244
bits to 247 bits and then it is given to decoder which converts back to 160
samples.
These 244 bits cannot be directly given to the decoder for the following
reasons:-
1. Decoder should know whether the frame what it is decoding is a good one or a
bad one so that the processing can be changed accordingly.. This is indicated by
BFI flag
2. Decoder should know whether the frame that it is decoding is a SID frame or
normal speech frame.. This is indicated by SID flag
3. Similarly decoder should have knowledge of Time alignment and this is
indicated by TAF flag.
These flags are calculated in the intermediate module and given to the decoder
and it decodes properly. This is used in TRAU of GSM wireless technology. In the
actual implementation, the intermediate module is implemented in the BTS on the
Uplink side and the flags that are generated are passed through abis
framing/deframing to TRAU where decoder alone is implemented. Decoder gets the
244 bits and also these flags and proper decoding takes place.
I hope i cleared your doubt
regards
parthu
Reply by Phil Frisbie, Jr.●April 1, 20052005-04-01
JH WANG wrote:
>
>
> Hi All,
>
> I am currently implementing the GSM EFR vocoders. There is one thing
> that i don't really understand.
>
> After encoding a file, i have to pass the file to another module so
> that the output format of that module will will allow my decoder to
> decode the file correctly. Why that module is not implemented on the
> encoder itself??
You must be using the reference code, right?
The output is modified to simulate communications loss and/or bit corruption.
Remember, GSM is designed for radio transmission so it must deal with loss or
corruption.
I converted the GSM HR to bit streams for Internet use, and it took MANY hours
to get it right!
> The additional module is to include bfi to each of
the frame of the
> encoded file.
Yes, I think that is to help simulate a radio receiver if I remember
correctly.
I am currently implementing the GSM EFR vocoders. There is one thing
that i don't really understand.
After encoding a file, i have to pass the file to another module so
that the output format of that module will will allow my decoder to
decode the file correctly. Why that module is not implemented on the
encoder itself??
The additional module is to include bfi to each of the frame of the
encoded file.