From: Andreas Stemmer <groovingandi_at_gmx.de>
Date: Mon, 20 Oct 2003 20:55:28 +0200

Hi BlueChip,

I don't want to start a flame on the best stereo mode. I was just interested
if I had something wrong in mind (I didn't look it up before I wrote my
message). I will comment on your links in the following if you don't mind.

BlueChip wrote:
> http://www.geocities.com/altbinariessoundsmusicclassical/vbr-e
> ncoding.htm#ch11 "Joint Stereo looks for signals that are identical
> in the left and right channel and if it finds any they are encoded as
> mono"

I found that one, too. Sounds like a non-techie explanation of encoding sum
and difference to me. Signals that are really completely identical in the
left and right channel, even for the short time of a frame, don't exist
(except you copy one channel digitally)

> with regard to the post on "you do not lose quality"
> === THIS IS NOT TRUE AT ALL ===
> Whatever you may read from the net - I can personally assure you
> otherwise. And Yes, I HAVE been put through the "blind test"
> procedure by two of my friends now.

What encoder did you use? I read that some commercial xing encoder had a
very poor joint stereo implementation in early times.

> http://www.mediatwins.com/FAQ/mp3_stereo_modes.htm
> "However, for some audio, Joint Stereo MS may create a 'flanging' or
> 'swishing' effect...

Yes, I read this one, too. It says "for SOME audio...". Of course, if you
have extremely wide stereo signals or two mono signals, you gain nothing by
using joint stereo. LAME automatically switches to normal stereo in theese
cases, but they are very rare. In normal audio, you gain bandwith or quality
by using joint stereo.

> http://www.mp3machine.com/discussion/messages/780.shtml
> quotes William Schelter, Nils Faerber, Alexander Leidinger, Oct 13,
> 2001: "Using mid/side stereo [MS] inappropriately can result in audible
> compression artifacts. Too much switching between mid/side and regular
> stereo can also sound bad."
> ...which was given as a definition of Joint Stereo

Again, I also found this site. It's not a definition of JS, it's just
describes the switching between JS and normal stereo with difficult signals.
The sentence which would come next in your citation is the real important
statement:
"To determine when to switch to mid/side stereo, LAME uses a much more
sophisticated algorithm than that described in the ISO documentation, and
thus is safe to use in joint stereo mode."

> http://www.iis.fraunhofer.de/amm/techinf/layer3/
> "Joint stereo coding takes advantage of the fact that both channels of a
> stereo channel pair contain far the same information. These stereophonic
> irrelevancies and redundancies are exploited to reduce the total
> bitrate." ...whatever that means

That's what I wrote.
If you have a normal stereo mp3 with 192kbps, you get 96kbps for each
channel. If you use joint stereo instead, you can use (I'm just guessing
numbers here) 160kbps for the mixed signal (sum of left and right). The
difference between left and right are very small, therefore 32kbps is enough
to encode it. The result is better quality for the main information.

> I cannot find defitions that agree with ANYTHING I have read so far,
> including my own post!!!

The technique you were describing could work, but I think it's not what is
known as joint stereo. All surround systems use the fact that deep
frequencies can't be located easily and have only one subwoofer. However,
the threshold frequency for this effect is in the range under 200Hz,
therefore you'll practically gain nothing.

Andreas Stemmer
Received on 2003-10-20