An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/speex-dev/attachments/20040715/b3253d77/attachment.htm -------------- next part -------------- ? Hi, I have some queries: I am using speex in Windows platform. 1. I am having some problem with the speex,for input 8KHz,8bits/sample mono stream i am getting output as 8KHz, 16bits/sample mono stream. I would like to know why 8bits/sample input stream is getting converted to 16bits/sample stream, will this have any effect on the sound clarity,if yes then to what extent and how do i slove this problem. 2. There is hissh sound in output stream. I would like to know how do i reduce the noise in the output and improve the sound clarity. 3. My requirement is 2kbps, how extent this low-bit rate will effect the sound clarity and how can i improve the sound clarity at lower bit rates. Thanks and Regards, Ram Narayan Rao.
>1. I am having some problem with the speex,for input 8KHz,8bits/samplemono stream i am getting output as 8KHz,>16bits/sample mono stream.In my app I'm using Speex codec. If I try to list all codecs for 8bits mode Speex is not present. It appears only for 16bits, so as I understand it's not for 8bits mode. Rgds, Ivan.
<003e01c46a4f$4da8df50$177e8cd5@massolit> Message-ID: <1089914436.15753.10.camel@idefix.homelinux.org> Why are people making a big thing out of this? It's not like converting to 16-bit linear is that hard, is it? Speex is a codec, not a sample rate/format conversion library. Jean-Marc Le jeu 15/07/2004 ? 05:36, Ivan Babikov a ?crit :> >1. I am having some problem with the speex,for input 8KHz,8bits/sample > mono stream i am getting output as 8KHz, > >16bits/sample mono stream. > > In my app I'm using Speex codec. If I try to list all codecs for 8bits mode > Speex is not present. It appears only for 16bits, so as I understand it's > not for 8bits mode. > > Rgds, Ivan. > > _______________________________________________ > Speex-dev mailing list > Speex-dev@xiph.org > http://lists.xiph.org/mailman/listinfo/speex-dev-- Jean-Marc Valin http://www.xiph.org/~jm/ LABORIUS Universit? de Sherbrooke, Qu?bec, Canada -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: Ceci est une partie de message =?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e=2E?Url : http://lists.xiph.org/pipermail/speex-dev/attachments/20040715/98bc9d6e/attachment.pgp
<0I0W00GUFF02LV@tid.hi.inet> <20040716084029.GA1732@kwaak.net> Message-ID: <1090343579.3194.6.camel@idefix.homelinux.org>> With a lot of fiddling and some good hints from jean-marc, I > actually got good reasonable good cancellation. > One of the biggest problem I had was that the dial-tone somehow > influenced the adaption-process that it continued singing along > for more than 30 seconds. By stopping the adaption process at the > moment that nothing comes into the mic (I am cancelling the echo > from the telephone line connected to the modem), it was greatly > improved.Did you try with the echo canceller in CVS (SVN actually), I think it should be better. Also, it's not when there's nothing in the mic that you should cancel adaptation, but when there's nothing in the speaker.> Second improvement was using auto-gain to get a constant "volume" > that needs to be cancelled.I think that's a bad idea, because it means you're changing the echo transfer function all the time and forcing the echo canceller to adapt all the time.> Third "improvement" was slowing down the adaption process. > Fourth "improvement" was adding "comfort noise" to give the users > a base noise to focus on. Without this noise, users tend to > increase the headset volume, which increases the amount of > feedback they hear, which makes them want to increase the volume > to hear the other side better, etc...Comfort noise always helps adaptation, so it's good.> BTW: I tend to increase the volume of the incoming signal by 2, > and divide the echosignal by 10 before I fed it into the > echocancel process.The echo cancellation is normalized internally, so what you're doing here has no effect. Jean-Marc -- Jean-Marc Valin <Jean-Marc.Valin@USherbrooke.ca> Universit? de Sherbrooke -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: Ceci est une partie de message =?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e?Url : http://lists.xiph.org/pipermail/speex-dev/attachments/20040720/a1eb2187/attachment.pgp
Hi, Somebody has tested succesfully the speex echo cancellator? Sometimes ago I post the same question, and it was very untested. Best regards, and thank you very much for your work. G.
<0I0W00GUFF02LV@tid.hi.inet> Message-ID: <20040716084029.GA1732@kwaak.net> On Thu, Jul 15, 2004 at 04:55:42PM +0200, Gustavo Garc?a Bernardo wrote:> Somebody has tested succesfully the speex echo cancellator? Sometimes ago I > post the same question, and it was very untested. > > Best regards, and thank you very much for your work.With a lot of fiddling and some good hints from jean-marc, I actually got good reasonable good cancellation. One of the biggest problem I had was that the dial-tone somehow influenced the adaption-process that it continued singing along for more than 30 seconds. By stopping the adaption process at the moment that nothing comes into the mic (I am cancelling the echo from the telephone line connected to the modem), it was greatly improved. Second improvement was using auto-gain to get a constant "volume" that needs to be cancelled. Third "improvement" was slowing down the adaption process. Fourth "improvement" was adding "comfort noise" to give the users a base noise to focus on. Without this noise, users tend to increase the headset volume, which increases the amount of feedback they hear, which makes them want to increase the volume to hear the other side better, etc... The problem (what I've noticed) is that the user is extremely perceptive for his own voice, so it really needs to be tuned down a lot. BTW: I tend to increase the volume of the incoming signal by 2, and divide the echosignal by 10 before I fed it into the echocancel process. (Still need to put this sofware on a webserver though... :-( )