Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: 320 cbr VS 256 VBR MP3 (Read 8461 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

320 cbr VS 256 VBR MP3

Maybe it was discussed before, but i could not find it...

For a non-technical mind like mine, it seems that 320 CBR MP3 (lame encoding) should provide the best possible quality in MP3's audio output.

But i read a discussion about joint-stereo vs real-stereo which made me wonder. Is it possible to achieve the same quality using 256 VBR? Or would it be possible that the 256 VBR encoding could provide even better audio output?

Now that more and more sites and making you pay your music by the chosen bitrate (allofmp3 for instance), the question is very interesting, i think.

If some experts can share their thoughts on the subject...

320 cbr VS 256 VBR MP3

Reply #2 –
and 320 vs 256 makes no difference quality-wise for that matter. at least audible difference

320 cbr VS 256 VBR MP3

Reply #5 –
Quote
Joint Stereo doesn't have anything to do with CBR or VBR.
[{POST_SNAPBACK}][/a]

hi. i understand that. what i was wondering is:
the same way joint-stereo  is allegedly better than real-stereo in some cases, since it can "free" some space in a file for valuable audio information... (ref: [a href="http://www.hydrogenaudio.org/forums/index.php?showtopic=26721]Listening tests discussion[/url]), maybe a similar phenomenon occur when u use VBR instead of CBR...
i guess dimt is right. no audible difference. but i'm curious.

320 cbr VS 256 VBR MP3

Reply #6 –
Quote
Quote
Joint Stereo doesn't have anything to do with CBR or VBR.
[{POST_SNAPBACK}][/a]

hi. i understand that. what i was wondering is:
the same way joint-stereo  is allegedly better than real-stereo in some cases, since it can "free" some space in a file for valuable audio information... (ref: [a href="http://www.hydrogenaudio.org/forums/index.php?showtopic=26721]Listening tests discussion[/url]), maybe a similar phenomenon occur when u use VBR instead of CBR...
i guess dimt is right. no audible difference. but i'm curious.
[a href="index.php?act=findpost&pid=246479"][{POST_SNAPBACK}][/a]


Check my last post.

320 cbr VS 256 VBR MP3

Reply #7 –
Quote
Maybe it was discussed before, but i could not find it...

For a non-technical mind like mine, it seems that 320 CBR MP3 (lame encoding) should provide the best possible quality in MP3's audio output.

But i read a discussion about joint-stereo vs real-stereo which made me wonder. Is it possible to achieve the same quality using 256 VBR? Or would it be possible that the 256 VBR encoding could provide even better audio output?

Now that more and more sites and making you pay your music by the chosen bitrate (allofmp3 for instance), the question is very interesting, i think.

If some experts can share their thoughts on the subject...
[a href="index.php?act=findpost&pid=246473"][{POST_SNAPBACK}][/a]


Fredhammersmith:  Here is an excerpt from a discussion a while back that may help out - rather long, but quite interesting:

rsadix wrote on 6-29-04:

Right from the lame --preset help it states
CBR 320kbps (highest quality possible from the --preset switches) The highest quality setting is cbr 320. As vbr approaches the limit of ultimate quality it becomes cbr 320. But vbr 320 isn't even a choice. As ABR 320 approaches the limit of ultimate quality and need it becomes cbr 320. From a quality perspecitve cbr 320 is not worse than abr 320. In fact as soon as abr has 1 frame that is not 320 it is less quality than cbr 320 albiet at the price of size. Small to negligible size.
We are told the pecking order is vbr best, abr less than vbr, cbr less than abr. But this is not true at all bit rates. Its certainly not true at the highest bit rates. So it would have to be true at lower bitrates, maybe the lower the bitrate the more true it is. Or is it? Where the cap of say cbr 192 is 192, vbr 192 has no cap and could possibly grab frames at 320 if it thinks it needs it. This is the best attribute of vbr or abr I believe. So what happens to a sharp attacking or smooth expanding sound that has part of it at one bit rate and another at a different bit rate. Is the integrity of the sound maintained across the frame transistion. Playing single notes on a piano is one thing, but .... all that time tweaking settings for the ever non replaceable rarewares 3.90.3, the blatant disastisfaction with every release of lame after 3.90.3 until possibly 3.96; all that time spent tweaking vbr... ahhh its probably all taken care of in the code somewhere; not to worry. Just have faith in --preset standard, or is is --preset standard -V1? Oh yeah and the "space savings" thats the other undeniable attribute of vbr; just like 55 saves lives. But what about bit rates of 112 or 96. People who record speech and transfer over the web, recording audio books etc. Those are mp3's. What vbr rate are those? Why they aren't vbr at all, there are no settings for vbr at those rates. That's where abr steps in again. The middle of the road guy between cbr and vbr. And since abr is variable like vbr it will always be smaller than cbr since cbr is wasting all that space recording silence it doesn't need. But wait at lower bit rates abr is gobbling up more space than cbr for quality. So when I'm trying to get my file size as small as possible abr is helping me the wrong way. The ear is good. At lower rates its challenging to hear the difference between cbr 48 and abr 48 and abr 48 is a larger file size. Hold on back to vbr. So vbr does have a sweet spot, and its obviously at the settings of medium - 128, standard - 192, extreme - 256, and insane at 320... wait thats cbr again. Lets go ahead and get Joint Stereo in the mix here. If you look at the distribution of Mid/Side Stereo vs Stereo in vbr files of different average rates you will notice that the --preset medium level files are mostly mid/side and much less pure stereo (depends somewhat on the music, but not as much when its dense like rock and roll). At the --preset standard level it becomes much more dependent on music but it swings from 50/50 to still mostly mid/side. However at the higher --preset extreme level its quite the opposite with many files having over 90% of their blocks on the pure stereo side. As we witness the convergence towards cbr pure stereo goodness. Along the same line; long blocks and short blocks also come into play. Their impact is not quite as dramatic. If its quiet there are short blocks if there is anything going on at all there are long blocks. In fact long blocks have it more than 90% of the time no matter what the bit rate.

So from a purely compatability perspective we would choose long blocks, as most of them are long anyway and changing block size might be an issue. Would there be a compression time bonus as well for not having to think about the long and short of it all - probably negligible. Being interested in higher quality sound, and witnessing that more than 90% of frames are pure stereo at higher bit rates wether vbr, abr or cbr we would probably choose pure stereo over mid/side stereo in our ideal compatible goal of not changing back and forth and just picking one. Possible compression time bonus. Its really a negligible size or quality difference at the higher rates anyway. And the last one is really a no brainer, variable rates are a significant contributing factor to incompatibility, granted incompatibility is small; so it has to be cbr. So thats cbr, stereo, and long blocks. For compatability anyway.

Viewing it from this harsh compatibility requirements perspective, there must be a flaw in size and/or quality. Lets start with size. Lets take a 3.5 minute song thats about a 36 meg wav file. cbr 320 is about 8.4 meg, cbr 256 6.7 meg, vbr 256 6.7 meg and vbr 128 4.7 meg. So it ranges from 13-23% of the original size. Or more practically and equally equated quality wise from vbr 192 to vbr or cbr 256 15-18%. Are we going to split hairs here over the size advantage of vbr. Well the last thing has got to be quality of sound. Well that's not something I can put numbers to right here and prove either way. Its certainly subjective but we do have some agreed upon sound samples to test with and ABX, however I'll take the Pepsi challenge with my
--preset cbr 256 -m s
settings over any vbr standard or vbr extreme file any day.
heck I'll even get dangersous and throw in a --noshort for a
--preset cbr 256 -m s --noshort
to boot.

Test it yourself and see if you can tell the difference. I have trouble hearing the difference.

I guess my point is, I just don't like seeing people ragging on cbr so much and people defending joint stereo so much. I think around the 256 bit rate vbr, abr and cbr are all very good quality and people are mainly listening to stereo even though their settings are for joint stereo. Lame is choosing stereo and long blocks except for the short lived quieter beginning and endings. So with indistinguishable quality, similar size and greater compatibility why poo poo cbr stereo. And why make lame spend resources figuring out the same thing over and over again.

So sure if you are trying to be ultra aggressive and save 1 meg per song and sacrifice a little quality go for --preset medium; its good. I'm busting my ass and I've almost got my 20 gig IPOD up to 7 gig. Heck I sacrifice some bandwidth and go for --preset voice (abr 56) versus cbr 48 but I'll wonder why when I can't play it on a friends player sometime down the road and dont have the original handy.

So don't get me wrong. I love lame. 3.90.3 and 3.96. I even had an affair with 3.93.1 and 3.95.1. I'm down with all mp3's at >192k quality compressed with lame; cbr, abr and vbr. I do wish more people would use lame and the default settings. So much unique info out there is recorded with non lame based sub acceptable junk its easy to see how mp3 takes a quality hit that it doesn't deserve. I'm glad the vbr issues of old are all but extinct in newer mp3 softwares and devices. I love mp3's and I love EAC. MP3 has still got everyone beet for sheer widespread borg domination. And while I'm rarely completely satisfied with downloaded mp3's (Think goodness for .shn and .flac) I'll keep making my own. I'll just be laughing/crying when that first in-car DVD player comes out that plays mp3's and I can put all the songs I have on a single DVD and it won't play some of them for vbr, joint stereo or some other lame reason. :bad005:

Just another note:  this topic has been widely debated and i think it's personal preference on what settings you ulimately use and depends on factors such as what you will be doing with the encoded clips...ex:  for backup purposes, to pay in a portable device, to play on a PC or home stereo, etc.

320 cbr VS 256 VBR MP3

Reply #8 –
Thanks dll2hell!
Very interesting.

Different sound systems, different results. In my system, i suffer a bit with 192 VBR MP3 files, the voices loose their purity and the spacial distribution is not very good.

I think i'll test the last Bjork at 320 CBR and 256 VBR.

Thanks again!

320 cbr VS 256 VBR MP3

Reply #9 –
Quote
rsadix wrote on 6-29-04:

Right from the lame --preset help it states
CBR 320kbps (highest quality possible from the --preset switches) The highest quality setting is cbr 320. As vbr approaches the limit of ultimate quality it becomes cbr 320. But vbr 320 isn't even a choice. As ABR 320 approaches the limit of ultimate quality and need it becomes cbr 320. From a quality perspecitve cbr 320 is not worse than abr 320. In fact as soon as abr has 1 frame that is not 320 it is less quality than cbr 320 albiet at the price of size. Small to negligible size.
[...]


For those interested, see http://www.digital-inn.de/showthread.php?t=25973 for the complete thread. (The Gecko over at digital-inn isn't me.) I am pretty certain that rsadix will loose his Pepsi challenge for example on castanets.wav.

@fredhammersmith

To answer your question: with mp3, 320 cbr is allways better than any vbr setting because 320 is the upper size limit an mp3 frame can be. You can not spend any more bits than 320 cbr. Any vbr setting will also produce frames that have less than 320kbps. If the vbr algorithm is good, then it will only scale down so there will be no audible difference between the 320 frame and whatever bitrate is chosen by the vbr algorithm. In this case, the 320cbr file and the (lower in bitrate) vbr file will sound exactly the same and you have conserved some space (assuming you are using settings which target transparency like lame --preset standard). If an imperfect vbr algorithm (all of them are imperfect) uses too low a bitrate, then - and only then - will the vbr file sound worse than the 320 cbr file.

Again, this is true only for cbr 320. A ~192 vbr file may possibly be better than a 256 cbr file, because on critical passages it can also use a 320kbps frame.

edit: please also see this post about stereo/joint-stereo: http://www.hydrogenaudio.org/forums/index....showtopic=28141