I've been an avid fan of 192 kbps CBR for a while now... it used to be 128 until my audiophile boyfriend started teasing me about the quality of CDs I made him. Lately I've been reading about ABR and VBR, too, and most (though not all) of what I have read says they provide better quality and file size than CBR, and they're split 50/50 over which is better, VBR or ABR. I really don't understand all the technical ways that some of these sites judge what it is "better", though.

I definitely like the idea of smaller file sizes, if ABR and VBR really are smaller. As far as quality goes, could someone please explain to me the differences in these three formats in as simple a way as possible? I can pick 96 kbps and poorer 128 kbps rips out of a line up, but anything higher is beyond me. I'm getting the feeling that I may be doing my music collection an injustice by making it all 192 CBR, but I'm not sure what other route to take. If anyone has any advice for me, I'd greatly appreciate it. Thanks so much for your time!

Edit: If I do go with VBR, what should my range be to get files that would be, on average, around the equivalent of 192 kbps CBR?

I use CBR. It is more compatible, more predictable and more straightforward.

Any time I do lame for music, I do 192kbps Constant Bit Rate, regular Stereo (not joint or what not). The simpler, the better, the less hassle.

I've listened carefully with headphones and find 192kbps constant sounds just as good as any of the variations, at least to my ears.

Thanks for the quick response! How have you found that CBR compares size-wise with VBR and ABR?

It is larger in file size, but with storage being so cheap, it seems to have a minimal impact.

I'd rather drop to 160kbps and stay CBR than go VBR, if space became a concern.

I find Fast Fraunhoffer (the one in Music Match 4.5 and on) sounds just as good to me at 160kbps as 192kbps does. But with LAME, I find I need to get to 192kbps to have the same quality level as the 160kbps Fast Fraunhoffer.

MP3 players are getting cheaper and store more, and DVD/CD's are way inexpensive, so I just stick with CBR cause there is less chance something will go wrong.

Ok, thanks so much for all the input, BeardedKirklander! I really appreciate it!

ABR is a variant of the VBR method of file encoding. When I encode files to MP3, I have always chosen VBR as it allows the encoder to choose the optimum bitrate for that particular section of music & avoids potential quality problems, though filesize as a result is unpredicatable. ABR gives most of the benifits of VBR encoding but with more predicatable final file size.

This provides a good explanation of it's benefits over CBR encoding:

http://wiki.hydrogenaudio.org/index.php?title=ABR

No sweat.

Another thing I have noticed, especially in long audio book files, is that when it's VBR instead of CBR, on some MP3 players, it seems to affect how they are able to read position / time left when you are in fast-forward / skip mode.

Anyway, good luck with the stuff.

In terms of quality, generally VBR > ABR > CBR at the same approximate size and bitrate.

CBR only has a limited "bit reservoir" to allow limited borrowing of bits from easier passages to assist with difficult ones.

VBR uses the acoustic model and quality level to determine the amount of bits used and their distribution. Encoders with sub-optimum acoustic models procuced miserable VBR results, clouding the early reputation of VBR.

ABR is a compromise, using a VBR-style distribution of bitrate, but subject to an overrall target bitrate, rater tha an overall target quality irrespective of bitrate.

As soon as you depart from CBR, you lose the exact correlation beteween data and time, usually resulting in problem with capacity estaimates at the very least.

The popular preset modes of LAME encoder, alt-preset standard and alt-preset extreme (actually, they changed to just "preset", and finally were absorbed into the VBR modes some time ago), are example of well tuned VBR with quality considerably better than a capacity-equivalent CBR.



Basically, for maximum quality in the space avaialble, or minimum space for whatever quality you consider acceptable - VBR.
If your player or software has any issues with VBR , USE CBR

There are few good applications for ABR... anything incompatible with VBR is likely to have problems with ABR, and if targeting a particular size, selecting & testing VBR is better.

Matth, thanks very much for the info!

Hi,

When I started on this quest a few years ago, I came across Chris Myden's website and his best mp3 guide.

The differiences I have from the guide is that I use the final version of LAME 3.97. Download

And I use the 'Additional Command Line Options' -V2 --vbr-new
(In EAC, click on EAC -> Compression Options and then the External Compression tab)

The result ? HQ MP3s.
As a note they all work on my IPOD, Car MP3 player, and audio CD's fine.

BeardedKirklander, you are making a mistake using Stereo rather than Joint Stereo. Joint Stereo is more efficient. Or to put it more clearly, Stereo is wasteful and hurts the sound quality.

Well, you are entitled to your opinion, but I'm perfectly happy with what I have and am not going to be changing from standard stereo anytime soon, or anytime in the distant future, I would think.

Compatibility and consistency matter to me more than some minute tweak in sound quality, if there even is one.

So don't worry about me. Just focus on what makes you happy, ok?

Joint stereo is a method to save some bandwidth by encoding certain parts of the spectrum in mono (i.e. only once) for which the human ear has no directional hearing. These are very low and very high tones.

The bandwidth is saved by recording a wider sum channel and a narrower difference channel, where the difference channel does not contain these spectral components.

This works very well and produces excellent quality at 128 Kbit/s for most pieces of music. The Fraunhofer codecs, for example, use this method.

However, there is one drawback. Some music contains sounds that are deliberately delayed or phase shifted. Such effect boxes are called "flanger", "phaser" and the like. If you encode such music in joint stereo, you will have bad cancelling effects where the high tones appear and disappear all the time, destroying the good original sound. One old example is the accompanying guitar in Paul Simon's "Mrs. Robinson".

Other encoders, like Lame or BladeEnc, record both stereo channels entirely separately. They are free of these distortions. However, to reach the same overall quality, they need some more bits, i.e. at least 160 Kbit/s.

Thus you could try Fraunhofer at 128 Kbit/s first, then listen for any distortions and, if you hear any, abandon the compressed music and compress again with Lame or BladeEnc at 160 Kbit/s or more. But, as memory becomes ever cheaper, you might as well use Lame or any other good encoder with separate stereo and variable bit rate encoding from the start.

Another way out of the dilemma is to use MPEG-4 AAC, which simply has a better compression algorithm and reliably produces excellent quality at 128 Kbit/s or even at 96.

To state that "joint stereo" is better than "Stereo" is false, it's not. Many times it may save space by making certain parts "mono" (i.e. changing it and assume you cannot hear the difference) but to say that it is better is ignorant.

Without entering all the details you guys just have been abble to explain above , I fully agree that "Stereo" gives the best final results .
Thank You , 123LTY , for having this situation clarified .

Different MP3 encoders use different terminology for the stereo modes, but for the LAME encoder "Joint Stereo" is the best stereo mode because it intelligently decides on when to use Left/Right Stereo, Mid/Side Stereo (and Intensity Stereo for low bitrates).

The normal stereo you know from Compact Discs, LPs, Casette Tapes etc, use two completely separate channels for Left and Right. For LAME this corresponds to L/R Stereo.

Instead of encoding each channel separately, you can instead encode the two channels as a "Mid" channel M=(L+R)/2 and a "Side" channel S=(L-R)/2. You can then calculate L and R as: L=(M+S)/2 , R=(M-S)/2. If you're using infinite precision those formulas are exact.

Most audio have Left and Right channels that are more similar than dissimilar, so the M(id) channel will be louder than the S(ide) channel most of the time, so since you don't have infinite precision when using MP3 ecoding, it makes sense to allocate more bits to the M channel and less bits to the S channel, or in other words it makes sense to take advantage of the fact that the L and R channels are not completely independent of each other.

The LAME encoder Joint Stereo mode will decide when it's best to encode using L/R Stereo and when it's better to use M/S Stereo (and for very low bitrates it can use Intensity Stereo as well). This is why it's called Joint Stereo and it is the best Stereo mode for the LAME encoder, because it either uses fewer bits with the same sound/stereo quality or uses the same number of bits with an improved sound quality or both.

This is not just a theoretical result, but has been confirmed by blind ABX listening tests.

You can read more about it on the HydrogenAudio forums, e.g. in this thread.

Other MP3 encoders may be different in how the stereo modes work and in the terminology used.

But see - you can avoid ALL of this hassle by just sticking with NORMAL STEREO.

I like life simple, and a bit more disk space is no big deal to me in this age where storage is cheap.

There's no hassle - it all happens automatically, and Joint Stereo is the default and recommended setting for the LAME encoder.

Even at the maximum bitrate for MP3 (320 kbps) you will get better sound quality using Joint Stereo instead of "normal stereo" (forced L/R Stereo) - unless you have a high-end Hi-Fi system and very good hearing, you probably won't be able to hear the difference, however.

There is no hassle with using Joint Stereo for the LAME MP3 encoder - you need to actively use non-default options in order to avoid it, and you would lose quality or get bigger files or both.

If you select the auto mode "-m a" in LAME

JS mode will normally improve quality/bitrate at medium bitrates, by using a mid/side transform where there is significant correlation, and L/R where there is not. There are modes which force mid/side at all times, but there are few occasions where they would be the right choice, and the switching point problems of old encoders are largely resolved.

The VBR "presets" tend to use JS, more for a potential space saving than for quality.

One thing though, on an underpowered system, encoding to "full stereo" (unconditionally using L/R frames) is faster than JS, and encoding higher bitrate is no slower (maybe faster) than lower bitrates, since the encoder is working harder to caculate the JS decision point.

Bigger files, no problem, but I don't see how you can get lower quality using the normal stereo option than joint stereo. That just doesn't make sense.

Giving the (LAME) encoder the option of encoding M/S Stereo instead of forcing it to encode L/R Stereo all the time, makes it possible for the encoder to optimize the number of bits allocated to making the original audio sound "good enough". This can then either be used to lower the amount of bits used thus decreasing file size while maintaining the subjective audio quality, or it can be used to increase subjective audio quality while maintaining file size.

You can think of it using an extreme example where there is no stereo information at all in the original audio. If you force an encoder to use L/R Stereo, it will effectively encode the L and R channels separately thus using twice the amount of bits needed to encode just one channel, or if file size is maintained it will have only half the amount of bits available per channel.

If you instead let the encoder decide how to encode the audio, it can detect that there is no stereo information, and encode in M/S Stereo with zero bits allocated to the S(ide) channel because the S channel is silent.

Real music is of course not that extreme.

again - space savings - I totally understand and that makes sense.

Taking space savings totally out of the equation, in terms of quality, I don't see how anything could be better than the original full stereo signal. You get all the musical data of both channels preserved, minus the high and low frequencies taken away by the actual MP3 compression.

If you encode audio to the mp3 format using the LAME encoder and forced L/R encoding, you will have 160 kbps available for encoding each of the L and R channels (half of 320 kbps).

If you OTOH let the encoder use whatever stereo mode (L/R or M/S) is best suited per audio frame, then you will have more than 160 kbps availabe for the M channel for some frames where there is correlation between the L and R channels and less than 160 kbps is needed to encode the stereo information in the S channel.

So even at 320 kbps you can get better audio quality by letting the encoder use the Joint Stereo format (mixed L/R and M/S).

Try to confuse my simple mind with all the technobabble you want, there is NO WAY that Joint Stereo will sound as good as TRUE NORMAL STEREO.

With normal stereo, all the data of both streams are still there. Not true with Joint Stereo.

With Normal Stereo, the LEFT channel is preserved. With Normal Stereo the RIGHT channel is preserved. Each is left in-tact, all the subtleties and nuances are captured and there are no signal oddities introduced that were not in the original recording.

I do not want or care about any of the space savings offerred by Joint Stereo. I want both stereo signals preserved independently, and that is what Normal Stereo gives you.

I like Constant bit rate. I like Normal Stereo. I like maximum compatibility with the least hassle. I don't see why space optimization on my files should matter to others if it does not matter to me.

I don't see how I am missing out on anything by using Constant bit rate and Normal Stereo. I'm happy, my playback devices are happy, life is good.

I have no interest in saving additional space by doing it any other way. I am not interested in sound quality per byte ratios and such. I just like it the way I like it and if I'm happy, people should be happy for me and move on.

Just because you didn't understand my explanation, doesn't mean that I was wrong and you are right.

No. Because mp3 encoding is lossy, all the information is *not* there regardless of encoding format.

You can do what you want with your ABR/CBR/VBR and stereo modes, but you are not getting the best possible quality/size of your mp3 files.

I will stop trying to convince you now.

I am not getting the best QUALITY TO SIZE ratio. That's fine.

But I don't CARE about the SIZE, you see.

I want NORMAL STEREO and I don't care about any size advantage by this joint stereo thing.

Thank you for stop trying to convince me. I appreciate that.