In the last Speex (http://speex.sourceforge.net) version (0.8.0), I added a new 42.2 kbps wideband mode which I'd qualify as the first non-catastrophic mode for music. I'd like to have feedback on music quality using that mode, let's say on a scale between "You call that music?" and "Almost as good as MP3". To try that mode, use the --quality 10 option *without* using --vbr. The file needs to be sampled at 16 kHz. The idea of this mode is to be able to encode occasional music segments in a mostly-voice recording (e.g. Internet radio). So what are the opinions?
BTW, Speex source code is available here (http://sourceforge.net/project/showfiles.php?group_id=46651) and there are Windows binaries on RareWares (http://www.inf.ufpr.br/~rja00/others.html).
I noticed that a 22 KHz, mono source is really comprensible, even for music.
Try to use Soundprobe (http://www.soundprobe.com/)'s expander & normalization before compressing... the sound can benefit from them !!!
I noticed that a 22 KHz, mono source is really comprensible, even for music.
Try to use Soundprobe (http://www.soundprobe.com/)'s expander & normalization before compressing... the sound can benefit from them !!!
I think if you compare an ogg created using '-q -1', downmix to mono and resample to 22050, you will get output at approx 16kbps which sounds somewhat better than the speex equivalent. At least it does to my ears!!
I think if you compare an ogg created using '-q -1', downmix to mono and resample to 22050, you will get output at approx 16kbps which sounds somewhat better than the speex equivalent. At least it does to my ears!!
You mean on music Ogg at 16 kbps sounds similar to Speex at 42.2 kbps? I consider that not bad since Speex has never been designed for music? Can you tell whether most of the artifacts in Speex are in the lower half-band or higher? (this is useful because both are encoded separately after a QMF).
In the last Speex (http://speex.sourceforge.net) version (0.8.0), I added a new 42.2 kbps wideband mode which I'd qualify as the first non-catastrophic mode for music. I'd like to have feedback on music quality using that mode, let's say on a scale between "You call that music?" and "Almost as good as MP3". To try that mode, use the --quality 10 option *without* using --vbr. The file needs to be sampled at 16 kHz. The idea of this mode is to be able to encode occasional music segments in a mostly-voice recording (e.g. Internet radio). So what are the opinions?
BTW, Speex source code is available here (http://sourceforge.net/project/showfiles.php?group_id=46651) and there are Windows binaries on RareWares (http://www.inf.ufpr.br/~rja00/others.html).
hey valin i would also want to ask that when i use command line speexenc and it decodes the .wav file and convert it into ogg container, and below that it also mentions that speex is not valid for 44.1khz but it still sampling..................