Speex dev - Aug 2004 - Videoconferencing with speex and jabber

Hello Jean-Marc,

Regarding your question - a little background on the various choices,
Fair Quality is 15kbps Speex narrowband, Good Quality is 24kbps Speex
narrowband, High Quality is 32kbps ADPCM from HawkVoice and Toll Quality
is 64kbps PCM.

We also have 10 other codecs within the product including GSM, LPC and
others that we currently do not expose.

To us, quality is a combination of perceived voice quality and latency.
Celp based codecs sound great when the tolerance range for the recording
input is within the range of human voice.  However, they tend to degrade
when people use low end microphones that introduce lots of ambient
noise.  We have found that at least half of people use low-end or table
microphones and quality of course suffers.  ADPCM and PCM are much
better in this regard because of the way they operate.  They are also
better with latency (of course) because there is less coding/decoding
going on.

This isn't a knock against Speex, as it works great at the bitrate and
we are very happy with it's performance vs. other voice coders like GSM
and are grateful for your efforts!

We authored the current Delphi header translation and DLL stuff for
Speex that is being used by lots of people.  While our header supports
Speex wideband, we have never been able to get wideband to work.

We may try another crack at getting it to operate with wideband in Speex
since you say it is so much more clean/better.
>Just curious, any reason you're not using Speex for high quality (at
18.2 and 24.6 kbps, I can't distinguish the coded from the original and
15 kbps is pretty close)? I'm also not sure I see the idea of narrowband
PCM either, since you can get better quality at lower bit-rates by
encoding in wideband (16 kHz).

        Jean-Marc


-- 
Jean-Marc Valin, M.Sc.A., ing. jr.
LABORIUS (http://www.gel.usherb.ca/laborius)
Université de Sherbrooke, Québec, Canada

<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.

> Regarding your question - a little background on the various choices,
> Fair Quality is 15kbps Speex narrowband, Good Quality is 24kbps Speex
> narrowband, High Quality is 32kbps ADPCM from HawkVoice and Toll Quality
> is 64kbps PCM.
Wow, these are pretty high bit-rates. Though I have never tested your 32
kbps ADPCM codec, my guess is that Speex 24.6 kbps probably has a higher
quality than an ADPCM codec at 32 kbps.
> To us, quality is a combination of perceived voice quality and latency.
> Celp based codecs sound great when the tolerance range for the recording
> input is within the range of human voice.  However, they tend to degrade
> when people use low end microphones that introduce lots of ambient
> noise.  We have found that at least half of people use low-end or table
> microphones and quality of course suffers.  ADPCM and PCM are much
> better in this regard because of the way they operate.  They are also
> better with latency (of course) because there is less coding/decoding
> going on.
If you're concerned with low-end microphones, I think you should try the
denoiser that's currently in the 1.1.x branch (try 1.1 because later
versions include minor changes in the API). 
> This isn't a knock against Speex, as it works great at the bitrate and
> we are very happy with it's performance vs. other voice coders like GSM
> and are grateful for your efforts!
I'm mostly trying to be informative here. I'm well aware that Speex is
not the best codec out there. 
> We authored the current Delphi header translation and DLL stuff for
> Speex that is being used by lots of people.  While our header supports
> Speex wideband, we have never been able to get wideband to work.
Can you tell me what the problem was? It shouldn't be harder to get
wideband to work than narrowband.
> We may try another crack at getting it to operate with wideband in Speex
> since you say it is so much more clean/better.
Definitely. Wideband and narrowband is day and night. At 27.8 kbps (and
possibly lower), you can get much better quality than anything you can
do with narrowband (see the samples on the website).

        Jean-Marc


-- 
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/20031118/7274beb2/signature-0001.pgp

Le mar 18/11/2003 à 17:39, Allen Drennan a écrit :
> Speaking of video conferencing in conjunction with Speex, we are
> currently beta testing a solution we developed that offers multi-point,
> multi-party video and audio using the Speex engine for voice.
> 
> http://www.wiredred.com/downloads/ecsetup.exe
> 
> The fair and good audio settings are Speex narrowband, high quality is
> HawkVoice and toll quality is PCM.
Just curious, any reason you're not using Speex for high quality (at
18.2 and 24.6 kbps, I can't distinguish the coded from the original and
15 kbps is pretty close)? I'm also not sure I see the idea of narrowband
PCM either, since you can get better quality at lower bit-rates by
encoding in wideband (16 kHz).

        Jean-Marc


-- 
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/20031118/b3c0574b/signature-0001.pgp