Question about VBR vs. 320kbps
Jul 17, 2008 at 9:58 PM Thread Starter Post #1 of 40

robenco18

Head-Fier
Joined
Jul 2, 2008
Posts
74
Likes
0
Hi,

I know what VBR is but when I have a choice between a file that is encoded at 320kbps or say 217kbps VBR which one is better? Is VBR always better or is 193kbps VBR not as good as 320kbps but 217kbps VBR is better than 320kbps. It isn't as clear cut for me as 320kbps vs. 256kbps is obviously. Should it be? Thank you.
 
Jul 18, 2008 at 5:07 AM Post #3 of 40
Well, 320kbps in theory will always be better than VBR 0. It's wasteful, and every single second of your track will be at 320kbps, regardless of whether it's needed or not. In practice, VBR is simply more efficient, and the resulting file is completely transparent. Then again, as is said, just try both and see if you can ABX them (foobar has a handy ABX tester built in). I'd be very surprised if you could actually pass the test, because AFAIK LAME VBR 0 is transparent.
 
Jul 18, 2008 at 6:09 AM Post #4 of 40
Depends what you look for, in the meaning of "better".
* Sound quality -> Then a 320kbps CBR file will most probably be "better" than a 217kbps VBR one.
* File size (for portable use) -> Then a 217kbps will probably be "better", as it have almost the same sound on just 2/3 the file size.
 
Jul 18, 2008 at 7:52 AM Post #5 of 40
These mp3 encoders will not use more than 320 kb/s for any portion of time regardless of settings, so a 320 CBR is indeed the maximum quality you can get. That doesn't mean it's a particularly wise choice. Chances are that 320 kb/s would have the same artifacts that -V0 would (in the case that you can hear any artifacts in -V0). Where space is an issue, the VBR setting would probably make a lot more sense, and when space is not much of an issue one should be considering lossless options.
 
Jul 18, 2008 at 9:04 AM Post #6 of 40
How many times... people who claim VBR saves "unused" bits which appear in 320kb/s mp3 CBR:
1) listen carefully on a decent equipment
2) find out what is "bit resorvoir" and how it works
3) try to extract "unnecessary" bits from the 320kb/s mp3 file, I forgot the name of the executable which does that trick. In 320kb/s, especially FhG it finds no spare bits.
-V0 doesn't even reach the 256kb/s quality level - just because of the fact of lower bitrate. Don't believe in miracles that less bits will sound better.
 
Jul 18, 2008 at 9:21 AM Post #7 of 40
Quote:

Originally Posted by majkel /img/forum/go_quote.gif
How many times... people who claim VBR saves "unused" bits which appear in 320kb/s mp3 CBR:
1) listen carefully on a decent equipment
2) find out what is "bit resorvoir" and how it works
3) try to extract "unnecessary" bits from the 320kb/s mp3 file, I forgot the name of the executable which does that trick. In 320kb/s, especially FhG it finds no spare bits.
-V0 doesn't even reach the 256kb/s quality level - just because of the fact of lower bitrate. Don't believe in miracles that less bits will sound better.



That program you are talking about is mp3packer (or WinMP3Packer is my GUI wrapped around it).

And yes, a file at 320kbps converted to VBR by removing unused data will usually bigger than a -V0 VBR file.. which means the original VBR file had some sort of loss.
 
Jul 18, 2008 at 9:45 AM Post #8 of 40
Quote:

Originally Posted by majkel /img/forum/go_quote.gif
3) try to extract "unnecessary" bits from the 320kb/s mp3 file, I forgot the name of the executable which does that trick. In 320kb/s, especially FhG it finds no spare bits.


Not unused, no, but useful? Depends on the listener. If you can't hear them, it's pointless to keep them. -V0 gives the encoder a chance to discard more where it believes it can, but whether it works out depends on the accuracy of that decision.

The way CBR works (roughly), the encoder iteratively picks frequency bands to make less accurate, then checks with the psy model to see if it's done any audible damage to the sound yet. It stops when it finds an acceptable result at the bitrate you asked for (or picks the best it can do if it can't get it perfect -- no miracles at 128kbps!). If you ask for 320bkps, it will use every bit of it, without even considering the possibility it could have used less.

Quote:

Originally Posted by majkel /img/forum/go_quote.gif
-V0 doesn't even reach the 256kb/s quality level - just because of the fact of lower bitrate. Don't believe in miracles that less bits will sound better.


It's around 700kbps. If I had to pick doing that or using 128kbps mp3 for my music, I'd pick the mp3. Efficiency, and not just raw bitrate, matters.

Yes, thanks to the bit reservoir, 256kbps mp3 can effectively reach 320kbps... for a very short time, until the reservoir is depleted. It has a very small fixed maximum size. -V0 is free to encode every frame at 320kbps if it feels the need. I've seen it average well above 256kbps.

Then again, if you don't trust the VBR mode, 320kbps CBR should never sound worse. The worst risk is a potential waste of space.
 
Jul 19, 2008 at 9:48 AM Post #9 of 40
Quote:

Originally Posted by LnxPrgr3 /img/forum/go_quote.gif
Not unused, no, but useful? Depends on the listener. If you can't hear them, it's pointless to keep them. -V0 gives the encoder a chance to discard more where it believes it can, but whether it works out depends on the accuracy of that decision.


Have a look at this page MP3packer - TMB Wiki

There can be wasted space that can be removed in a completely lossless fashion from a 320kbps CBR file. But as LAME is progressing, less and less wasted space is appearing in the 320kbps files.
 
Jul 19, 2008 at 12:46 PM Post #10 of 40
Quote:

Originally Posted by majkel /img/forum/go_quote.gif
-V0 doesn't even reach the 256kb/s quality level - just because of the fact of lower bitrate.


LAME almost invariably uses a substantial number of 320kbps frames when it encodes at -V0.
 
Jul 19, 2008 at 2:43 PM Post #11 of 40
Quote:

Originally Posted by Febs /img/forum/go_quote.gif
LAME almost invariably uses a substantial number of 320kbps frames when it encodes at -V0.


This means nothing. 320kb/s frames appear even in the 128kb/s CBR mode, at least the FhG encoder uses them occasionaly. The higher the bitrate, the more offen it happens.
 
Jul 19, 2008 at 6:38 PM Post #12 of 40
-V0 vs. -b320
biggrin.gif


Hearing a difference between -V0 and -b320 isn't unheard of -- for people listening for artifacts in problem samples. If -V0 were universally transparent for all sounds anyone would ever want to encode, on all equipment, to all people, LAME developers probably wouldn't be toiling with public listening tests and years of tuning between versions.

Given that, not everyone will necessarily hear the same differences. People who hear artifacts in problem samples at -V0 can probably still use that setting for most of their music (unless they only listen to known problem samples).

All in all, the only way to be able to say what the differences between setting is to you, as many have said, is to try it out!
 
Jul 21, 2008 at 10:24 PM Post #13 of 40
Quote:

Originally Posted by robenco18 /img/forum/go_quote.gif
Hi,

I know what VBR is but when I have a choice between a file that is encoded at 320kbps or say 217kbps VBR which one is better? Is VBR always better or is 193kbps VBR not as good as 320kbps but 217kbps VBR is better than 320kbps. It isn't as clear cut for me as 320kbps vs. 256kbps is obviously. Should it be? Thank you.



VBR is simply more efficient at minimizing file size relative to a specific level of sound quality. The real question is to determine where your threshold of transparency lies. Beyond that, you will achieve no gains in sound quality but a substantial increase in the size of your files. The only reason to use lossy encoding is to save space and to achieve the maximum sound quality relative to the maximum space savings, use --preset standard or -V 2. If you find that the sound quality at this setting is lacking, you can increase stepwise, -V 1 or -V 0, toward the maximum sound quality relative to the minimum space savings, i.e --preset insane or -b 320 until you find your sweet spot.

Lame-chart-2.png
 
Jul 22, 2008 at 8:32 AM Post #14 of 40
This graphs are pure speculation. What do the quality steps mean? Between the -v2 and -v0 is a slight difference due to more aggressive low-pass filter, and it's audible in the treble region when you compare. However, when switching from -V0 to 320kb/s there is a substantial difference in the midrange distortion level and timbre completeness. Unfortunately, most of the budgetary audio equipment and DAPs doesn't even reach the quality level of 256kb/s CBR or -V0, and that's why people don't hear the difference. So, for me this graph is crap, because there is a meaningful difference for me between the 320kb/s CBR and everything below which has incomplete tonality.
 
Jul 22, 2008 at 9:55 AM Post #15 of 40
Quote:

Originally Posted by majkel /img/forum/go_quote.gif
This graphs are pure speculation. What do the quality steps mean? Between the -v2 and -v0 is a slight difference due to more aggressive low-pass filter, and it's audible in the treble region when you compare. However, when switching from -V0 to 320kb/s there is a substantial difference in the midrange distortion level and timbre completeness. Unfortunately, most of the budgetary audio equipment and DAPs doesn't even reach the quality level of 256kb/s CBR or -V0, and that's why people don't hear the difference. So, for me this graph is crap, because there is a meaningful difference for me between the 320kb/s CBR and everything below which has incomplete tonality.


Sounds like you found your sweet spot.
 

Users who are viewing this thread

Back
Top