IPB

Welcome Guest ( Log In | Register )

2 Pages V   1 2 >  
Reply to this topicStart new topic
Is 320 CBR the best MP3 sound quality?
TheGrimRipper
post Feb 13 2010, 11:01
Post #1





Group: Members
Posts: 30
Joined: 11-March 06
Member No.: 28391



Hi people.

Apols if this has been asked umpteen times but I could not get a clear view on this.

OK VBR is the recommended standard.

But, I've seen comment that it sounds better than 320 CBR. Which I'm a bit confused about.

Ignoring file size aspects, surely 320 CBR would potentially be the best sound quality?

Cheers.
Go to the top of the page
+Quote Post
punkrockdude
post Feb 13 2010, 11:12
Post #2





Group: Members
Posts: 256
Joined: 21-February 05
Member No.: 20022



I think some encoders have had 384 kbps as an option but I don't know if that was outside of any ISO/offical specs. Others will be more of an help.
Go to the top of the page
+Quote Post
shadowking
post Feb 13 2010, 11:14
Post #3





Group: Members
Posts: 1527
Joined: 31-January 04
Member No.: 11664



320 cbr is the highest 'mainstream' quality for mp3. You can have higher quality with freeformat upto 640k, but most decoders don't play it.


--------------------
Wavpack -b450s0.7
Go to the top of the page
+Quote Post
EagleScout1998
post Feb 13 2010, 11:14
Post #4





Group: Members
Posts: 275
Joined: 1-October 06
Member No.: 35820



"Sounds better" is a subjective term. VBR (variable bit rate) is better than CBR (constant bit rate) not because it "sounds better", but because it is more efficient.
Go to the top of the page
+Quote Post
TheGrimRipper
post Feb 13 2010, 11:17
Post #5





Group: Members
Posts: 30
Joined: 11-March 06
Member No.: 28391



Thanks for replies, yes I'm talking "mainstream".

I'm particularly interested in 320 CBR vs VBR V0.

Go to the top of the page
+Quote Post
TheGrimRipper
post Feb 13 2010, 11:22
Post #6





Group: Members
Posts: 30
Joined: 11-March 06
Member No.: 28391



QUOTE (EagleScout1998 @ Feb 13 2010, 10:14) *
"Sounds better" is a subjective term. VBR (variable bit rate) is better than CBR (constant bit rate) not because it "sounds better", but because it is more efficient.


Yes I fully concur that in terms of the file size/quality ratio, VBR is the place to be.

But ignoring file size, surely VBR could never sound better than 320 CBR. Assuming a version of LAME had been used that handled CBR well.
Go to the top of the page
+Quote Post
lvqcl
post Feb 13 2010, 11:35
Post #7





Group: Developer
Posts: 3411
Joined: 2-December 07
Member No.: 49183



Usually both CBR320 and V0 are transparent so they are equal in terms of sound quality.

There are tracks where CBR320 is better than V0.

But maybe there are also tracks where V0 is better than CBR320... LAME uses different compression algorithms for VBR and for CBR/ABR.
Go to the top of the page
+Quote Post
shadowking
post Feb 13 2010, 12:49
Post #8





Group: Members
Posts: 1527
Joined: 31-January 04
Member No.: 11664



V0 should never be better than b320. If it is its some fluke. On some rare samples 320 or --abr 287 -h give better performance than V0

For the most defensive strategy that has mainstream playback ; -b320 or very high bitrate ABR is the way to go.


--------------------
Wavpack -b450s0.7
Go to the top of the page
+Quote Post
psycho
post Feb 13 2010, 13:06
Post #9





Group: Members
Posts: 241
Joined: 14-October 05
Member No.: 25099



shadowking, so you're saying there are no problem samples, which affect CBR 320 kbps and don't affect VBR V0?


--------------------
lame -V 0
Go to the top of the page
+Quote Post
birdie
post Feb 13 2010, 14:25
Post #10





Group: Members
Posts: 108
Joined: 3-March 06
From: this planet
Member No.: 28235



If your HDD/storage media permits, store and get all music in lossless format.

There's no excuse to compress audio nowadays.
Go to the top of the page
+Quote Post
[JAZ]
post Feb 13 2010, 14:27
Post #11





Group: Members
Posts: 1785
Joined: 24-June 02
From: Catalunya(Spain)
Member No.: 2383



We are mixing here theory and practice.


The teory is clear that CBR320 cannot be better than VBR, because VBR uses less, or at most the same amount of bits to encode the same thing.


In practice, they are different beasts, like in the use of the bit reservoir or the difference in the way the analisys is applied (search for best within bitrate constraint versus search for required amount within quality constraints).

Go to the top of the page
+Quote Post
TheGrimRipper
post Feb 13 2010, 14:48
Post #12





Group: Members
Posts: 30
Joined: 11-March 06
Member No.: 28391



QUOTE ([JAZ] @ Feb 13 2010, 13:27) *

We are mixing here theory and practice.


The teory is clear that CBR320 cannot be better than VBR, because VBR uses less, or at most the same amount of bits to encode the same thing.


That's what's confusing me. I always assumed that because VBR V0 varies the bitrate up to 320, whereas CBR 320 sticks slavishly to 320, VBR V0 cannot sound better then CBR 320. As good as perhaps, in so far as not to be able to notice the difference. But better?

QUOTE ([JAZ] @ Feb 13 2010, 13:27) *

In practice, they are different beasts, like in the use of the bit reservoir or the difference in the way the analisys is applied (search for best within bitrate constraint versus search for required amount within quality constraints).


This is the interesting bit. Has LAME been so fine tuned to VBR that VBR V0 sounds better than CBR 320 even though VBR V0 uses less space (ergo is even more lossy)?

This post has been edited by TheGrimRipper: Feb 13 2010, 14:50
Go to the top of the page
+Quote Post
TheGrimRipper
post Feb 13 2010, 14:54
Post #13





Group: Members
Posts: 30
Joined: 11-March 06
Member No.: 28391



QUOTE (birdie @ Feb 13 2010, 13:25) *
If your HDD/storage media permits, store and get all music in lossless format.

There's no excuse to compress audio nowadays.


MP3 is still more generically supported than FLAC for example. And in environments such as in the car, lossless vs lossy becomes a lot less noticeable especially at 320. And my alpine head unit doesn't decode FLAC. smile.gif
Go to the top of the page
+Quote Post
birdie
post Feb 13 2010, 14:55
Post #14





Group: Members
Posts: 108
Joined: 3-March 06
From: this planet
Member No.: 28235



Absolutely most people won't tell a difference between 192KBit MP3 and CD audio, so choose whatever you feel is right for you and for your requirements.

Theoretically 320CBR can have the same quality as --preset extreme (the best VBR quality) in lame.

This post has been edited by birdie: Feb 13 2010, 14:55
Go to the top of the page
+Quote Post
birdie
post Feb 13 2010, 14:59
Post #15





Group: Members
Posts: 108
Joined: 3-March 06
From: this planet
Member No.: 28235



QUOTE (TheGrimRipper @ Feb 13 2010, 18:54) *
QUOTE (birdie @ Feb 13 2010, 13:25) *
If your HDD/storage media permits, store and get all music in lossless format.

There's no excuse to compress audio nowadays.


MP3 is still more generically supported than FLAC for example. And in environments such as in the car, lossless vs lossy becomes a lot less noticeable especially at 320. And my alpine head unit doesn't decode FLAC. smile.gif

Car audio is usually has so low quality, 192 VBR will be indistinguishable from CD audio - trust me.

Or don't trust me and go though a simple blind test - burn an audio CD with the same song burned say 10 times in audio CD quality and decompressed mp3 in random order.

This post has been edited by birdie: Feb 13 2010, 15:00
Go to the top of the page
+Quote Post
TheGrimRipper
post Feb 13 2010, 15:06
Post #16





Group: Members
Posts: 30
Joined: 11-March 06
Member No.: 28391



QUOTE (birdie @ Feb 13 2010, 13:59) *
Or don't trust me and go though a simple blind test - burn an audio CD with the same song burned say 10 times in audio CD quality and decompressed mp3 in random order.


You are speaking to the converted here. smile.gif

The main reason for this thread is that someone who seems to talk knowledgeably said that VBR V0 blew CBR 320 away in terms of quality.

And I find that statement unbelievable.

So I thought I'd come back here to the font of all knowledge on such things and ask if things had radically changed in the few years since I last visited. smile.gif
Go to the top of the page
+Quote Post
trout
post Feb 13 2010, 15:07
Post #17





Group: Members
Posts: 424
Joined: 26-March 09
Member No.: 68400



God kills a kitten every time you encode with CBR 320.

</sarcasm> rolleyes.gif
Go to the top of the page
+Quote Post
[JAZ]
post Feb 13 2010, 15:09
Post #18





Group: Members
Posts: 1785
Joined: 24-June 02
From: Catalunya(Spain)
Member No.: 2383




VBR may fail when it cannot see that it needs a higher bitrate than the one used, while CBR320 would be always using that bitrate (so innmune).
Then, there is also another thing related with the bit reservoir that can affect the distribution of bitrate in VBR, and as a consequence not have enough bitrate for a specific case*

CBR may fail in different cases. Even if it may seem ilogical, it may try to keep too much quality on some parts, and unexpectedly degrading others.

As said, VBR does not use the same procedures as CBR, and you can create a CBR file with the VBR algorithm (and with an attached VBR header) as was proposed once to research about a possible bug in CBR: http://www.hydrogenaudio.org/forums/index....st&p=583384



* say it decides it needs only a 32kbps frame, while the next doesn't have enough with a 320kbps one. If it had encoded the first one with 64kbps, it could have more space in the bit reservoir. Anyway, there are some limits that apply here too so I'm not sure how much of a difference can be in reality.


[Edit: removed first sentence. it didn't have sense]

This post has been edited by [JAZ]: Feb 13 2010, 15:11
Go to the top of the page
+Quote Post
TheGrimRipper
post Feb 13 2010, 15:20
Post #19





Group: Members
Posts: 30
Joined: 11-March 06
Member No.: 28391



Thanks, great link & very useful info.

I'm using LAME 3.97 at the moment for CBR, is it worth going to 3.98?

I'll likely give VBR a go to and do some blind testing. Is 3.98 the release to go with for VBR?

This post has been edited by db1989: Jul 29 2011, 12:57
Reason for edit: removing pointless full-quote
Go to the top of the page
+Quote Post
halb27
post Feb 13 2010, 15:50
Post #20





Group: Members
Posts: 2439
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



Yes, it's worth going to 3.98, especially when using VBR.

3.97 had inadequate problems with a certain kind of distortion called 'sandpaper noise problem' as well as with a certain trumpet sample. Moreover AFAIK there was tuning with 3.98 towards the more usual pre-echo problems.
So the quality of 3.98 is more homogeneously good than that of 3.97. This is more important when using VBR than when using ABR/CBR, as the VBR quality relies more heavily on the quality of the psy model and its implementation.



--------------------
lame3100m -V1 --insane-factor 0.75
Go to the top of the page
+Quote Post
/mnt
post Feb 13 2010, 16:28
Post #21





Group: Members
Posts: 697
Joined: 22-April 06
Member No.: 29877



I do have track that does sound worse with 320kbps then V0.

320kbps vs FLAC
CODE
foo_abx 1.3.4 report
foobar2000 v1.0
2010/02/13 14:55:54

File A: J:\Music\Lossless\Kraftwerk - The Man Machine\03. Metropolis.flac
File B: C:\Temp\Metropolis 320 CBR.mp3

14:55:54 : Test started.
14:56:26 : 01/01 50.0%
14:56:30 : 02/02 25.0%
14:56:34 : 03/03 12.5%
14:56:38 : 04/04 6.3%
14:56:42 : 05/05 3.1%
14:56:46 : 06/06 1.6%
14:56:49 : 07/07 0.8%
14:56:53 : 08/08 0.4%
14:56:57 : 09/09 0.2%
14:57:02 : 10/10 0.1%
14:57:07 : 11/11 0.0%
14:57:12 : 12/12 0.0%
14:57:17 : 13/13 0.0%
14:57:21 : 14/14 0.0%
14:57:25 : 15/15 0.0%
14:57:26 : Test finished.

----------
Total: 15/15 (0.0%)

Pre-echo throughout the start, but it's not annoying. Sadly at 0:09 there is a very obivous and annoying precho artifact.

320kbps vs V0
CODE
foo_abx 1.3.4 report
foobar2000 v1.0
2010/02/13 15:00:10

File A: C:\Temp\Metropolis 320 CBR.mp3
File B: C:\Temp\Metropolis V0.mp3

15:00:10 : Test started.
15:00:39 : 01/01 50.0%
15:00:42 : 02/02 25.0%
15:00:48 : 03/03 12.5%
15:00:52 : 04/04 6.3%
15:00:55 : 05/05 3.1%
15:01:02 : 06/06 1.6%
15:01:07 : 07/07 0.8%
15:01:10 : 08/08 0.4%
15:01:14 : 09/09 0.2%
15:01:19 : 10/10 0.1%
15:01:24 : 11/11 0.0%
15:01:30 : 12/12 0.0%
15:01:34 : 13/13 0.0%
15:01:38 : 14/14 0.0%
15:01:42 : 15/15 0.0%
15:01:45 : 16/16 0.0%
15:01:46 : Test finished.

----------
Total: 16/16 (0.0%)


At the start V0 does produce slightly rougher precho, but at 0:09 the 320kbps file stands out greatly. While the 0:09 mark on the V0 file lacks the annoying artifact.

QUOTE (TheGrimRipper @ Feb 13 2010, 15:20) *
I'm using LAME 3.97 at the moment for CBR, is it worth going to 3.98?


If you are already using 320kbps CBR on LAME 3.97 with no compat issues i would stick with it. Since LAME 3.98 has some regression issues with problem tracks, due to it using a smaller frame size to make it 100% ISO compliant.

This post has been edited by /mnt: Feb 13 2010, 16:31


--------------------
"I never thought I'd see this much candy in one mission!"
Go to the top of the page
+Quote Post
lvqcl
post Feb 13 2010, 18:35
Post #22





Group: Developer
Posts: 3411
Joined: 2-December 07
Member No.: 49183



So we have two opinions:

3.98 is better than 3.97:
QUOTE (halb27 @ Feb 13 2010, 17:50) *
3.97 had inadequate problems with a certain kind of distortion called 'sandpaper noise problem' as well as with a certain trumpet sample.

3.97 is better than 3.98:
QUOTE (/mnt @ Feb 13 2010, 18:28) *
If you are already using 320kbps CBR on LAME 3.97 with no compat issues i would stick with it. Since LAME 3.98 has some regression issues with problem tracks

Go to the top of the page
+Quote Post
Soap
post Feb 13 2010, 18:47
Post #23





Group: Members
Posts: 1016
Joined: 19-November 06
Member No.: 37767



I was under the impression the sandpaper noise issue was only in regards to VBR, is this correct?
Regardless, both statements are factually correct, as both are real issues not present in the other.

This post has been edited by db1989: Jul 29 2011, 12:57
Reason for edit: removing pointless full-quote


--------------------
Creature of habit.
Go to the top of the page
+Quote Post
timcupery
post Feb 13 2010, 19:18
Post #24





Group: Members
Posts: 780
Joined: 19-December 01
From: Tar Heel country
Member No.: 683



The reason that 3.98 regresses compared to 3.97, with some problem tracks, is because the Lame devs changed the handling of the bit reservoir. Basically, some outdated decoders had trouble with 320kbps frames that used too much bit reservoir from previous frames. Unfortunately, one such outdated decoder was in WMP, which is used by many people. I haven't been able to replicate this problem with WMP 11 or 12 (on XP and Win7, respectively), but I understand the Lame devs' goal that Lame mp3's play on anything that handles mp3.

So basically, 3.98 can't go as high of a bitrate in a single frame, as 3.97, and thus will struggle with a few problem samples.

As I understand it, the actual spec definition of mp3-format's bit reservoir is somewhat vague, so Lame 3.97 and prior could be read as spec-compliant, but so could the problematic WMP decoder.

But the tuning changes between 3.97 and 3.98 have still improved 3.98 for other issues.


--------------------
God kills a kitten every time you encode with CBR 320
Go to the top of the page
+Quote Post
lvqcl
post Feb 13 2010, 19:41
Post #25





Group: Developer
Posts: 3411
Joined: 2-December 07
Member No.: 49183



QUOTE
I haven't been able to replicate this problem with WMP 11 or 12 (on XP and Win7, respectively)

Try mplayer2.exe
Go to the top of the page
+Quote Post

2 Pages V   1 2 >
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 21st October 2014 - 13:47