Re: QDM2 progress report
Benjamin Larsson <[email protected]>
| Newsgroups | gmane.comp.video.xine.codec.devel |
|---|---|
| Message-ID | <[email protected]> |
Hi. Jim Leonard wrote: > I'm not sure I'm following you; you suspect that the data is getting > companded to (assuming a mono sample) an 8-bit mu-law format Correct. > before going through the second phase? I can't fathom why they would > do that (but then again you're the one RE'ing the scheme). Companding is used in most codecs I think. If you requantize PCM you can get down to 12bits/sample before the quantization noise gets to noticeble. If you compand it you can get it down to 11 bits/sample. But going from 16 to 8 bits is to harsh. I can only assume they did it because it was easy. But when doing encodings with MVP it is done, first I just thought they did it for low bitrates like 24kbit and 48kbit. That would make a little sense to reduce data when low on space, but even 128kbit uses mu-law companding. Thus rendering the encoder quite useless. /Benjamin ------------------------------------------------------- This SF.Net email is sponsored by: IBM Linux Tutorials Free Linux tutorial presented by Daniel Robbins, President and CEO of GenToo technologies. Learn everything from fundamentals to system administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click