[vorbis] joint stereo - advantages / when?

[Aleksandar Dovnikovic]

Hugo van der Merwe <hugovdm at mail.com> wrote:

> I've been postponing some of my encoding for when joint stereo gets
> implemented. The reason I've been doing this, is that I am under the
> impression this is the largest step in the quality/bitrate ratio that's
> left.

I think that's true. The saving that is achieved by using joint stereo
varies - if there is big stereo separation & difference between
audio channels then the gain is small/none, but in practice this
rarely happens so usually you can save ~30-40kbps by using
joint-stereo, and sometimes even more.

> Now I'm wondering if I am correct in thinking this. lame's
> documentation seems to imply it doesn't make much of a difference at
> higher than 128kbps (I quote: "To disable joint stereo encoding
> (slightly faster, but less quality at bitrates <= 128 kbps)"), but this
> could just be because one anyway doesn't hear much of a difference
> between 128 and 160 kbps?

It does make a difference - and LAME is a good example :-) From my
experience, using LAME at 192kbps you will get less artifacts when
you use joint-stereo instead of simple stereo. Also most test files
sound better at 256kbps JS then 256kbps S.

Bottom line: if joint-stereo is properly implemented (and LAME has
pretty good joint-stereo), then I don't see the point in using simple
stereo because you will only get bigger files and they won't sound
better... of course there is still probability that joint-stereo might
slip up somewhere but as I said, if the implementation of joint-stereo
is good then there is no worry. Only at high bitrates (depends on the
codec, but generally speaking ~256kbps and higher) you should/could
use simple stereo, because at such high bitrates there are enough bits
so there is no real need for savings (and you also avoid that small slip
up probability of joint-stereo).

Greetings,
Aleksandar

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.