192kbps stereo OR 128kbps joint stereo

which mostly better ?

Depens on encoder and settings used as well as on the encoded material.

but if other parameters are same?

I'd probably go with 192kbps stereo.

I'd choose 192kbps joint-stereo. Or -q6 Vorbis.

OK, OK. I can't hear artifacts very well, so I'd probably choose 128kbps joint-stereo mp3.

In terms of quality, I'll choose 192 stereo

I'd go for 192 kbps joint-stereo, because it usually allows the encodes to attain a higher (but probably inaudible at this bitrate) quality than 192 kbps with just true-stereo.

Edit: There are samples? I based my answer on a negative assumption, and with knowledge from the forum.

Well, the question was 192kbps stereo or 128kbps joint-stereo.

Just in case you didn't noticed... The 192 kbps sample is using simple stereo while the 128 kbps sample uses joint stereo - there is no 192 kbps JS.

Edit: bidz was faster.

If the material does not hahe a big difference in the elft/right channels i would go 128kbits joint stereo
If the channels are very different i would go 192 full stereo

If the L/R channels are 100% identical, a 128kbits joint stereo would give same quality as 256kbits full stereo.

Geez, the title on this thread should be "Fish hooks or corkscrews"....that's the only way you can describe the audible quality of those settings.

Umm... I think your statement is wrong. Joint Stereo encoding should select between Mid/Side Stereo and Simple Stereo (at least if the MP3 encoder is reliable) frames automatically, depending on the difference between the channels.
Both Joint Stereo and Simple Stereo should therefore reproduce the same stereophonic sound.

PS: Of course, if the Joint Stereo file also uses Intensity Stereo frames, a loss of the stereo-image will occur, resulting the Simple Stereo file to have a better quality.

Quick question about Joint Stereo. Encoding MP3s using the LAME build from the sticky --alt-preset extreme.

Should I keep it using JS or should I use stereo? Don't care about file size, quality is more of an issue for me.

edit:

Nm, found a good thread that had some great info about my question. Going to stick with JS

StanleyPain: With LAME, using joint stereo is always better quality wise. Especially so for you sinc you use --alt-preset extreme; this preset is already *extremely* conservative in choosing when to use joint stereo frames vs stereo frames.

Adding to sven_bent's comment, the 192 simple stereo sample is probably encoded with a higher lowpass than the 128, so unless the listener is sensitive to compression artifacts, it may be the better choice even with low stereo separation samples.

And yet you are missing the encoder, that 128js could be ruined if they used, say, old xing... And the 192 stereo is ruined, if they used, say, Blade...

Well, he did ask which one is *mostly* better, so this should be taken as general advice. The analysis would certainly apply to most commonly and currently available MP3 encoders.

He is in fact correct. In samples with high stereo separation, you're getting little bitrate savings benefit from the joint stereo option, because the joint stereo ends up setting most frames as simple stereo anyway. His identical-channels example is the most extreme consequence of this effect. To take the opposite extreme, in channels that are completely different (no correlation at all), 128 kbps joint stereo will be identical to 128 kbps simple stereo because js will always choose simple stereo frames--and both will be equally worse choices than 192 simple stereo.

All else being equal, including bitrate, properly-implemented joint stereo will give you the same stereophonic sound as simple stereo, but you'll also get less obvious compression artifacts the lower you go in bitrate, so it's always a better choice. Unfortunately, the point is he has to choose between one sample or the other in this case.

The first part is misleading and the second part is flat out wrong.

As I understand it, even when picking stereo, one channel can use bits leftover from the coding of another channel -- there doesn't have to be 100% separation of bit allocation.

The only case when you get a scenario like what you are describing (where 256kbps stereo wouldn't be any better than 128kbps js) is when you are using dual mono, which isn't the same thing as simple stereo.

I thought that was exactly what he wrote... isn't dual mono by definition L/R channels being 100% identical?

I think you're wrong Dib, sorry

A song with 100% identical channels would use 50% of the bitrate for left, 50% for right in a stereo OR dual-mono configuration, which means that a 128kbps file would use 64kbps on each channel.

A joint-stereo version of the same song would use 100% bitrate on the mid channel (which in this case is the same as both the L and R channels), and 0% on the side channel (since it would be empty, L-R=0 where L=R). Therefore you would effectively get 128kbps for each channel on playback.

In practice, it's hard to answer the question because practically all songs fall someplace in the middle. And of course, different encoders do a better job of JS than others.

.

I can't believe we're discussing it!

Given that you said "mostly", I assume you don't want to know about all the exceptions to the rule. So the answer is simple: 192kbps Stereo.



As for the discussion: 128kbps JS is not as good as 256kbps S for mono sources. This was discussed on the old r3mix forum, and some testing was done - the differences were quite apparent. There's mention of this at the very bottom of this page:

http://www.ff123.net/mono.html

Cheers,
David.

Out of curiosity, I decided to do the same type of test (source with 2 identical L/R) at the same lowpass (17.5 kHz). One using --preset cbr 128 (uses joint stereo by default) and the other using --preset cbr 256 -m s. Clear difference between the two. Clear difference between -mj -b128 --lowpass 17.5 and -ms -b256 --lowpass 17.5 as well. Maybe joint stereo needs more bits in overhead that simple stereo does for this case, or an inherent problem with using joint stereo in extreme cases like this one.

Regardless, theory did not translate very well to practice in this case. Oh well.

EDIT: spelling

whats better? apples or berries?

In terms of quality-to-filesize ratio, 128kbps CBR JS is better.

Recommendation: Use "--preset cbr 128" or "--preset standard" instead.

threads like this will confuse new users and strenghten the old myths about simple stereo usually equaling to better quality(which is WRONG).

Inform yourself about the presets, and let them automatically do the decision on which options to use. They usually result in much better quality, than setting the options yourself - and they are easier to use.

A good encoder usually can better decide about which options are best, than the user - so you should trust it and don't care about stuff like JS or SS - because its to YOUR advantage, your files will usually have better quality with lame presets.

edit: maybe the lame-devs shall rename joint-stereo to intelligent-stereo...... maybe then clueless users who fiddle with settings they dont understand, just becaues the -names- of certain options sound better, will stop to use simple stereo. Or even better: remove the simple stereo switch completely from stable builds.

I kind of assumed he had a choice between downloading (or sharing?) two files, encoded using these parameters.

Hopefully any amount of time spent reading this board would dispel any myths that either setting is optimal. Still, it's amazing how many people don't read the FAQ (or anything else!) before posting.


btw, I wouldn't recommend "--preset cbr 128" - I'd recommend "--preset 128", because abr can be dramatically better around this bitrate. Of course --alt-preset standard is the way to go in place of 192kbps stereo.

Cheers,
David.