Hi,
Speex is getting close to beta4, which I'd like to be feature-complete
(or as close as possible). That's why I'd like to ask if anyone here has
needs for a feature that hasn't been implemented yet. If so, please let
me know.
For those interested, here's what's going to be in beta4. First, the VBR
code has been greatly improved and now works good with wideband too.
There are also improvements in the packet concealment code and the VAD
(Voice Activity Detection). It's also possible to have the VAD without
VBR, so that Speex does comfort noise generation during silence, but
otherwise constant bit-rate. Last thing, the 2.15 kbps mode now has a
better (OK, less ugly) quality.
Jean-Marc
--
Jean-Marc Valin, M.Sc.A.
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: 242 bytes
Desc: signature.asc
Url :
http://lists.xiph.org/pipermail/speex-dev/attachments/20021213/7229dec5/signature.pgp
On Fri, Dec 13, 2002 at 01:21:17AM -0500, Jean-Marc Valin wrote:> Speex is getting close to beta4, which I'd like to be feature-complete > (or as close as possible). That's why I'd like to ask if anyone here has > needs for a feature that hasn't been implemented yet. If so, please let > me know.Hi Jean, I've one small request - an option on speexenc that allows you to specify a speex file to append to, allowing you to concatenate streams without losing quality by decoding & encoding. Ideally, it would: * Read the sampling rate, etc, from the file and use those. * Toggle the end of stream marker on the last frame of the file * Continue appending new frames from thereon. And probably ignore/conflict with settings which would change the whole file (eg, comments, change in input format (unless an Ogg/speex stream can have different audio formats in the same file?), narrow/wideband, etc). This idea only just sprung to mind when you offered - I'm currently using a program derived from speexenc/speexdec to simply join two or more speex files together - but I'll happily write up a patch against CVS in the next 2 or 3 days and send it to you, if you or others think this useful (and not just excess bloat). Kind regards, Bernard. -- Bernard Blackham bernard at blackham dot com dot au Australian Linux Technical Conference 2003: http://www.linux.conf.au/ -------------- next part -------------- A non-text attachment was scrubbed... Name: part Type: application/pgp-signature Size: 190 bytes Desc: not available Url : http://lists.xiph.org/pipermail/speex-dev/attachments/20021213/7753006c/part.pgp
> I was wondering if you could add a check to ensure that memory is actually > allocated during the nb_encoder_init and sb_encoder_init functions. We have > been looking at using Speex on a DSP and noticed that if we didn't allocate > a large enough heap space memory segment that the DSP would crash. I would > recommend something like: > if (!st->stack) fprintf(stderr,"ERROR could not allocate memory for stack");I guess I should do that on all other malloc'ed variables too.> Also, is it necessary to allocate 20000 floats for stack space. Have you > ever profiled the amount of stack space consumed? We tried a few values for > the stack size and checked for stability. This isn't the most rigorous > check... However, we did reduce the stack size down to 5000 in both nb and > sb_encoder_init functions and we were able to successfully encode at quality > 6, wideband, complexity 3. On a DSP memory can be scarce, so we may need to > cut this down some more if possible.I noticed that, in CVS the stack size for the encoder is now down to 4000 (2000 for the decoder). This size works for any quality/complexity setting. For lower complexity settings though it might be possible to set a lower value, but I haven't found any good way to handle that so the size is set to the maximum size possible (because complexity can be changed dynamically). I'd be happy to receive ideas on better ways of handling that... Jean-Marc -- Jean-Marc Valin, M.Sc.A. 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: 242 bytes Desc: signature.asc Url : http://lists.xiph.org/pipermail/speex-dev/attachments/20021213/98a0f3a1/signature-0001.pgp
Jean-Marc,
I was wondering if you could add a check to ensure that memory is
actually
allocated during the nb_encoder_init and sb_encoder_init functions. We have
been looking at using Speex on a DSP and noticed that if we didn't allocate
a large enough heap space memory segment that the DSP would crash. I would
recommend something like:
if (!st->stack) fprintf(stderr,"ERROR could not allocate memory for
stack");
Also, is it necessary to allocate 20000 floats for stack space. Have
you
ever profiled the amount of stack space consumed? We tried a few values for
the stack size and checked for stability. This isn't the most rigorous
check... However, we did reduce the stack size down to 5000 in both nb and
sb_encoder_init functions and we were able to successfully encode at quality
6, wideband, complexity 3. On a DSP memory can be scarce, so we may need to
cut this down some more if possible.
Thanks,
Trevor Yensen
AvalonDSP Consulting Inc
<p><p>-----Original Message-----
From: owner-speex-dev@xiph.org [mailto:owner-speex-dev@xiph.org]On
Behalf Of Jean-Marc Valin
Sent: Friday, December 13, 2002 1:21 AM
To: speex
Subject: [speex-dev] Speex wishlist
<p>Hi,
Speex is getting close to beta4, which I'd like to be feature-complete
(or as close as possible). That's why I'd like to ask if anyone here has
needs for a feature that hasn't been implemented yet. If so, please let
me know.
For those interested, here's what's going to be in beta4. First, the VBR
code has been greatly improved and now works good with wideband too.
There are also improvements in the packet concealment code and the VAD
(Voice Activity Detection). It's also possible to have the VAD without
VBR, so that Speex does comfort noise generation during silence, but
otherwise constant bit-rate. Last thing, the 2.15 kbps mode now has a
better (OK, less ugly) quality.
Jean-Marc
--
Jean-Marc Valin, M.Sc.A.
LABORIUS (http://www.gel.usherb.ca/laborius)
Université de Sherbrooke, Québec, Canada
--- >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.
This is not exacty a core feature request, but ... Has there been anyone working on an MS ACM wrapper for Speex? (Ie, to allow it to be used by NetMeeting). If not, it's a project I'd be happy to take up for the next couple of weeks. I'd really like to replace MS's incompatible GSM 6.10 implementation w/ Speex (and the audio quality is so much better!) Regards, Brian -- Brian Teague PGP Key: 06B9AC63 on keyserver.net Baker '04 Fingerprint: 0C9A C8E2 7254 60F4 4342 Rice University 6383 6DC9 9783 06B9 AC63 Give me wisdom as my only lover On the lonely road Break your searing silence And light my journey home -- "Byzantium" --- >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.
Le ven 13/12/2002 à 12:43, Brian Teague a écrit :> This is not exacty a core feature request, but ... > > Has there been anyone working on an MS ACM wrapper for Speex? (Ie, to > allow it to be used by NetMeeting). If not, it's a project I'd be happy > to take up for the next couple of weeks. I'd really like to replace > MS's incompatible GSM 6.10 implementation w/ Speex (and the audio > quality is so much better!)I don't use windows myself, but this would definitely be a good thing to do. If I'm not mistaken, this could allow NetMeeting to communicate with GnomeMeeting (which already supports Speex) using Speex. If you want to do that, I can assist you regarding the Speex API, but I know nothing about Windows programming. Jean-Marc -- Jean-Marc Valin, M.Sc.A. 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: 242 bytes Desc: signature.asc Url : http://lists.xiph.org/pipermail/speex-dev/attachments/20021213/e7dae1d3/signature-0001.pgp
Hello Bernard, Friday, December 13, 2002, 7:22:54 AM, you wrote: Bernard> I've one small request - an option on speexenc that allows you to Bernard> specify a speex file to append to, allowing you to concatenate Bernard> streams without losing quality by decoding & encoding. Ideally, it Bernard> would: But you can `cat speex1.ogg speex2.ogg> unionspeex.og` and still have a valid gapless file. Correct me if I'm wrong. -- Best regards, Alejandro Belluscio <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.