<div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div><div>
</div></div></blockquote><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Adjusting the code to reduce/merge bands is someplace on my todo list,<br>
but you can try it easily enough by adjusting bark_freq in<br>
libcelt/modes.c and change the 12000 to 11024 and you should be able<br>
to get the full bandpass of your 22050 signal.</blockquote><div><br>
Thanks a lot. This worked perfectly.<br>
</div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Would you really need support for 22050 at 24kbit/sec if CELT could<br>
produce identical results at 44100 24kbit/sec? If your interest is<br>
just in operating at lower bitrates then there is a lot we can achieve<br>
by running at full rate and low-passing internally.</blockquote><div><br>Yes,
I am looking for lower bitrates, optimized for 22 kHz signals. Ideally
I'd like to have good quality for speech at about 40 kbit/sec and get
better quality than Speex. So far, by doing your hack and removing the
two highest bands I get completely transparent quality at 62 kbit/sec. <br><br>The results I get with CELT at 42 kbit/sec are also pretty good,
but the highest bands get some noise in the 8-11 kHz frequency range. I am
going to see what I can do, and try to optimize the band allocation or
other parts of the codec. For comparison, AAC-LC achieves transparent quality for speech at about 48 kbit/sec.<br>
<br></div></div>
Marc.<br>