Speex dev - Aug 2004 - status report on the Windows CE (PocketPC) ACM codec

Hello,
I had tried to run smapleenc.c and sampledec.c
from the manual (don't confuse it with the encoder and decoter command line
utils ; the code ther is different).
the final decoded audio file is with a very poor quality audio
when using it with 16 bit PCM raw audio input .

Did anybody tried these samples with success ?
Or should I add something ?
regards
JOHN Z.

_________________________________________________________________
The new MSN 8: advanced junk mail protection and 2 months FREE* 
http://join.msn.com/?page=features/junkmail

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'speex-dev-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

Hello,

just a brief report on where I am going with that Windows CE ACM Speex codec.

* I managed to compile the GSM610 sample codec by Visual C++ Embedded
   for the ARM processor. This is set up as a DLL project in the IDE. Replacing
   the implementation with the existing Win32 code should be straightforward
now.

* I wrote an install/setup procedure for the codec (works fine)

* I built a cross-compiler for target "arm-newlib-coff" on the CygWin
platform
   (which runs on Windows and emulates a unix shell and related GNU apps)

* The cross compiler can build speex-1.1.3 and generates a .lib file which I
   hope to link into the codec. The advantage of this approach is that I do not
   need to modify the Speex distribution much.

<p>Now here is where I have problems and not yet found a solution:

* the function names exported in the .lib created by gcc are not compatible
  with the function name decorations that the VisualC++ embedded linker expects.
  I keep getting unresolved external symbol messages.
  A __stdcall function for example would have to look like this (the number
  obviously indicating the number of bytes passed as arguments on the stack)
  _function@4
  However gcc will not emit these function names. I tried to force this to
  __cdecl type functions which would have names like _function, but so far
  without success.

* Loading resource strings from the codec DLL seems to fail. The codec
  name turns out blank on the PocketPC, however that is just a minor flaw
  and could be fixed by hardcoding those strings.

*Future conflicts might arise from the 2 different stdlib implementation.
  The gcc code will use the open-source "newlib" while the Visual C++
compiled
  code will rely on Microsoft's libc.

<p>Well, with some luck I will be able to compile a working codec, but the
linking
issue is a real showstopper right now.

Is there a way to post-process a .lib file to alter its export symbols?

Christian

<p>--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'speex-dev-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

These sample encoder are meant to be used for speech sampled at 8 kHz,
mono. Is that what you were using?

        Jean-Marc

Le dim 04/01/2004 à 09:31, john paul a écrit :
> Hello,
> I had tried to run smapleenc.c and sampledec.c
> from the manual (don't confuse it with the encoder and decoter command
line
> utils ; the code ther is different).
> the final decoded audio file is with a very poor quality audio
> when using it with 16 bit PCM raw audio input .
> 
> Did anybody tried these samples with success ?
> Or should I add something ?
> regards
> JOHN Z.
> 
> _________________________________________________________________
> The new MSN 8: advanced junk mail protection and 2 months FREE* 
> http://join.msn.com/?page=features/junkmail
> 
> --- >8 ----
> List archives:  http://www.xiph.org/archives/
> Ogg project homepage: http://www.xiph.org/ogg/
> To unsubscribe from this list, send a message to
'speex-dev-request@xiph.org'
> containing only the word 'unsubscribe' in the body.  No subject is
needed.
> Unsubscribe messages sent to the list will be ignored/filtered.
-- 
Jean-Marc Valin, M.Sc.A., ing. jr.
LABORIUS (http://www.gel.usherb.ca/laborius)
Université de Sherbrooke, Québec, Canada


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 190 bytes
Desc: Ceci est une partie de message numériquement signée.
Url :
http://lists.xiph.org/pipermail/speex-dev/attachments/20040107/1b674b2c/signature-0001.pgp

Christian Buchner wrote:
> Now here is where I have problems and not yet found a solution:
I know very little about this, so take everything that follows with a 
large pinch of salt.
> * the function names exported in the .lib created by gcc are not compatible
>   with the function name decorations that the VisualC++ embedded linker
expects.
Which version of gcc are you using? I know that only recently has gcc 
made itself compatible with the C++ name mangling standard, whatever 
that is. I have no idea if MSVC++ is compatible with it.
> *Future conflicts might arise from the 2 different stdlib implementation.
>   The gcc code will use the open-source "newlib" while the Visual
C++ compiled
>   code will rely on Microsoft's libc.
In general, you have to compile everything with the same compiler to get 
it to link correctly.
> Is there a way to post-process a .lib file to alter its export symbols?
I believe it's not just a matter of symbol names, but also of calling 
convention - so one lot of code will put arguments on the stack in a 
different way to that expected by the other lot. This can't be fixed by 
post-processing.

Maybe the gcc lists are a better place to ask.

Gerv
--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'speex-dev-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.