On Friday, October 18, 2002, at 09:19 AM, Bill Warner wrote:
> from the document that you provided I think this is more of a lossless
> compression only between the stereo channels. Rather than throwing out
> stereo samples that it would normally deem unnecessary it would keep
> them. The over all sound sample is still lossy.
>
> Although I don't know much about ogg, and that is just my
> interpretation
> of the doc you sent to the list.
This was my interpretation as well. It goes through a process ("stereo
coupling" or something similar) where it examines the left and the
right channel of the small sound sample it is working on. If it deems
that they are close enough in tone, then it will encode only one music
channel and record the relative volume offset that should be applied
when decoding. From my experience, it determines that about 90% of the
music I listen to can be combined in this way.
As someone else mentioned, FLAC or something else would be required for
lossless compression. Keep in mind, though, that blind listening tests
have repeatedly revealed that people cannot tell the difference between
a high-quality Ogg and the original source (even the medium quality
setting is almost impossible to distinguish). Lossless encoding is
only required when the music is going to be edited ("remixing", etc.)
since it is likely that the music will be encoded multiple times.
- --
Voltage Spike
,,,
(. .)
- --ooO-(_)-Ooo--
(text/plain)