On 30 January 2013 05:42, Albert Cahalan <[email protected]> wrote:
> This is not a speech codec. It is a white noise generator.
>
> Has anybody tested other non-x86?
>
> If little-endian ARM works, then endianness is likely at fault.
> If it fails, then char being unsigned by default is likely at fault.
>
> I've tested (using c2enc and c2dec) on ARM-LE and it worked as expected.
c2enc does this:
while(fread(buf, sizeof(short), nsam, fin) == (size_t)nsam) {
codec2_encode(codec2, bits, buf);
where "buf" is an array of shorts. It looks to me like this would fail if
you were reading one of the raw sample files (which are 16 bit little
endian.)
The output of c2dec would also be 16 bit big endian.
I think all that would be needed to test this is to convert the raw sample
file (with sox) to big endian, pass it through c2enc and c2dec and remember
that the output is going to be 16 bit big endian. Actually fixing the
issue will be more work of course...
Dan MD1CLV
------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_jan
_______________________________________________
Freetel-codec2 mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freetel-codec2
