Joint Stereo vs Stereo (?)

[liquidfluidity]

I have currently started ripping my cd collection to mp3 and FLAC. I am using DB Poweramp and ripping at a cbr of 320kbps. My question is - since I am ripping at such a high bitrate , will I gain a better quality rip using "joint stereo" or "normal stereo"? I started ripping using joint stereo and , in the car , I can actually tell a difference in the separation. I am using the Lame encoder and am seeing many different opinions regarding the stereo ripping choices versus quality. One being that "joint" decreases the separation between channels but is better quality wise for lower bitrates - the other being that stereo is better for higher bitrates. I'm only interested in quality - time or space is not an issue. My next step I guess will be to rip something using the different stereo methods and see if there really is a difference. What's your guys opinions outs there? I'm looking at ripping probably 2000 discs so I don't want to get half way through and decide to start over.

 

[Procyon12]

Lame's Joint stereo is lossless in its stereo separation.
therefore, technically speaking, the same exact audio quality will take up less space on disk.
and then, theoretically, audio ripped at the same bitrate will be better in joint stereo...

 

[Sycraft]

Well if the sort of thing bothers you, you can look at using FLAC or other lossless compression instead. Since it is 100% mathematically lossless it is the same as the orignal CD.

Joint stereo should be what you want as the encoder will then decide on a per frame basis if it needs to use mid-side mode or go for two separate channels. Stereo mode forces the encode to be two 160k mono streams.

Really at 320k though it should make no audible difference. You are going beyond the limits of human perception. If you really want to test it, you need to set up a blind test to make sure your opinions aren't influencing things.

Like I said if it is just a mental thing with you, then use FLAC and be happy as that is as high a quality as you can possibly get since it is bit-accurate to the original. I personally store all my CD music in FLAC (and my DVD-A music in MLP) because storage space is cheap as hell so there is no reason to worry about compressing it with lossy compression. I do compress it losslessly since there is also no reason not to do that (and in the case of DVD-A it is MLP on disc anyhow).

 

[liquidfluidity]

I am using MP3 in the car but am also ripping FLAC. I read initially that joint was the way to go but like I said , I thought things didn't quite sound right in the car. I did a little more investigating earlier and that's where i found the varying opinions between the 2. Here's one quote that made me start rethinking what i was doing and maybe it was relevant to the "difference" that I was seemingly "hearing".

"In a nutshell, Stereo means that each channel is encoded independently and requires a higher bitrate to maintain good quality, whereas Joint Stereo uses a clever mathematical algorithm to sum the difference of both channels along with the mono content and stores it as a single data stream.

The latter works well with lower bitrates (usually less than 160kbps) but may not be a true stereo reproduction." Kind of a "joint" con.

But then I find quotes like this from Hydrogenaudio:

"Simple stereo is the most straightforward method of coding a stereo signal: each channel is treated as a completely separate entity. This can be inefficient and may adversely impact quality (as compared to other modes) when both channels contain nearly identical signals (i.e., are mono or nearly so). "

Obviously this one supports the pro opinion for joint stereo. When I started ripping I was all about the numbers but when I actually started listening to what I was doing , it made me question my whole program set up which is why I'm here. I wish the head unit in my car supported FLAC!

 

[Fryguy8]

If you have an environment where you can easily tell the difference between the 2, then recreate that environment, play both types, decide which you like the sound of better, and use that.

 

[liquidfluidity]

That's the plan. I just thought I'd get some opinions and see what others had experienced. Thanks for the replies thus far.

 

[Sycraft]

If you are wondering how they work on a technical level it is as follows:

Music on a CD is stored in L/R format meaning there is a left channel and a right channel stored as a separate, monaural stream. They are then sent to the respective speakers on playback. Easy to do and understand. However there's another way of storing the same data called M/S, mid-side. Here one channel stores the information that is the same between left and right, the mid channel, and another stores the different, the side channel. The math is simple M = L+R, S = L-R. Works well and is mathematically lossless. You convert to M/S mode for various reason, in audio production is can be done for processing, like to alter the stereo field. In the case of compression, it is done to get better compression.

Now as this applies to MP3 every frame can be compressed wither L/R or M/S (there's another mode but it only applies to low bitrates). In a stereo frame, L/R, two monaural channels, are used. In a joint stereo frame the encoder converts the stream to M/S and encodes it that way. The advantage is it can spend more bits on the mid signal than the side one. This is useful because most music has far more in common between the channels than different.

When you choose stereo, it forces L/R encoding all the time. When you choose joint stereo mode the encoder chooses on a per frame basis to use L/R or M/S encoding depending on the material and available bitrate.

 

[liquidfluidity]

I have everything stored on my desktop so I think I'm going throw my headphones on and listen to some CDs and some rips and see/hear the differences if available.

Sycraft , according to your explanation , in it's essence , joint stereo DOES actually change the original data by separating and mixing and then separating the two channels again whereas normal stereo contains 2 completely separate tracks of data which are handled individually - which is how the original source is actually made.

I understand that I am probably making this a lot more than it really is and that it's going to come down to opinion on my part. Thanks guys.

 

[Sycraft]

The act of going L/R to M/S is completely lossless. FLAC does it too. They are mathematically equivalent and you can convert freely form one to the other. The only change is in how the bits are allocated.

 

[liquidfluidity]

But it's actually different than the original correct? If joint is able to decide which way to go, what dictates this and wouldn't any change create a difference whether it's lossless or not? I would think the separation woud change however slightly.

I'm a musician also so maybe my hearing is a little more sensitive to change. Maybe it's just me. I'll do some testing over the next few days and post back.

 

[Sycraft]

No there's no difference is L/R and M/S when you are talking uncompressed data, You can go back and forth freely, and lose no information. It isn't a question of acoustics, but of math. when I say "L+R" I mean literally the samples in the channels are added together.

It is similar to the difference between RGB and HSL (or YUV) in pictures or video. Just different ways of storing the same information and you can convert back and forth no problems.

Now in terms of MP3, you lose data no matter what. That's the point of lossy compression. It is just a matter of how it is done. With a stereo frame, you'll keep 100% of the stereo separation (more or less) but more bits can't be allocated to the detail of common signals. With joint stereo less bits can be used on separation, more on detail as needed.

Over all, joint stereo should give a superior sound quality at a given bitrate since musical signals do have a lot in common between the channels. It isn't like stereo magically gets more bits or something.