Pollinators Secret Mixter Signups
skip
Home » Forums » DIY » VBR vs. CBR

VBR vs. CBR

fourstones
.
permalink   Tue, Oct 9, 2007 @ 2:08 PM
I’ve had to write this up a few times in email so I figured I would post it here for a permanent reference.

When encoding to mp3 there are 2 kinds of encoding methods: constant (CBR) and variable (VBR). Some encoders (esp. those embedded in DAWs) only know how to do CBR.

The magic of VBR (if I understand it properly) is that the encoding rate gets higher at the parts of your mix that need it and lower where it can get away with it. I don’t know 100% how it “knows” or even if that’s right, what I do know is that all my mixes sound much, much better using VBR than an equivalent size MP3 using CBR.

If you’re on Windows I can HIGHLY recommend the WinLAME encoder

On the ‘VBR’ settings the encoding (to my ears) is freaking brilliant, yet compact.

Or if you’re on the MAC you get the same encoder integrated into iTunes here:
http://www.blacktree.com/ap...

Again, make sure to use the ‘VBR’ setting.
duckett
.
permalink   Tue, Oct 9, 2007 @ 5:17 PM
Probably a stupid question, but here goes: would the highest VBR settings available in ACID Music Studio sound as good as the WinLAME encoding? All my stuff has pretty much been CBR; I figured VBR was lossier, and since my uploads mostly come under 10MB even with using 320, I never worried about it. If I can have equivalent sound for a smaller file size, I’m all for it!
 
.
permalink   fourstones Tue, Oct 9, 2007 @ 5:32 PM
LAME is free so do a comparison.

(I think ACID uses Fraunhofer?)
Loveshadow
.
permalink   Wed, Oct 10, 2007 @ 4:11 PM
This is a useful topic and my advice is try it.

In reality there should be no percievable audio difference in a VBR or CBR encoded file.

If anything the CBR file should be what it said it is, a constant level of quality where as a VBR will do just that , vary it.

I encode CBR as i dont want to have a machine decide when to trade quality for file size if i can. The 10 meg limit here is a luxury for mixes and the vbr standard is only required if you were over six minutes at 320k.

The codec differences were primarily designed for video and encoding for DVD. The picture information present in a single frame varies from frame to frame. A static shot of a wall does not need a lot of information to screen it. But a close up of a basketball match in action would.

A VBR encode would reduce the output bit rate when the wall was on and boost it when the baskeball match was playing. That way you can cram more quality digital info onto a 4.7 gig dvd.

There are I frames to consider in video aswell but thats another subject.

Using that info a CBR rate is giving the chosen level of quality for every frame of your mix.

And 5 mins a 320k will get you in at 10 meg. Anything over that might need a VBR codec which will do as it does in video giving no percieved quality loss but giving a smaller file size.

The one thing i have said before is that all this quality is going to waste if your mix is not sounding good in the first place.

One thing to add is that Lame is just the code and you need a front end application to encode. Thats fact confuses some people when they are searching an encoder havent checked Fourstones link but will guess its a complete package.
 
.
permalink   fourstones Wed, Oct 10, 2007 @ 5:13 PM
er, I just to have say as an admin on the site: the goal is not to get as close to 10MB as possible. The truth is, if every upload was in fact that large we would have to drastically reduce the limit.

Along with server space there are bandwidth considerations: bitrate is related to bit stream rate and the biggest drag is when your player stops midway for ‘buffering’.

So for the web, the smallest file with the best sound you can squeeze out is what you’re aiming for.
 
.
permalink   Loveshadow Thu, Oct 11, 2007 @ 3:43 AM
Ok well thats another matter and thanks for pointing it out.

With that in mind it is probally better to experiment with VBR codecs and smaller files sizes then.

We dont want this place falling over. :-)
 
.
permalink   duckett Thu, Oct 11, 2007 @ 6:16 AM
I’m going with VBR from now on; I had the same thoughts as to quality, but after listening to both CBR and VBR versions of a mix,
the only perceptual difference I noticed was the VBR seemed a little “cleaner” and a little thinner, (maybe?) making more use of the brain’s ability to “fill in” the missing data. Smaller uploads mean I hopefully continue to be tolerated as I struggle to suck less. Even if my remixes aren’t works of genius, I’ll always try for the best audio quality because A) my stuff’s usually too short to worry about, and B) it’s an attempt at honesty. I’ll never claim the reason for a track’s lack of love is “Oh, well, if you heard the 48kHz WAV of it, you’d change your mind”. Like you said, a bodykit and custom paint can’t turn a Fiesta into a Mustang… I just hope I’m not DJ Fiesta ;)
 
.
permalink   celionati Mon, Feb 4, 2008 @ 9:49 PM
Quote: Loveshadow In reality there should be no percievable audio difference in a VBR or CBR encoded file.

If anything the CBR file should be what it said it is, a constant level of quality where as a VBR will do just that , vary it.

Well, this isn’t correct. Bit rate and perceived quality are not at all the same thing, and a well designed VBR method can produce better perceived quality than a CBR file of the same size. The obvious example is that silence can be encoded with very few bits, just by giving the length of the silence. There is no loss of fidelity encoding silence that way. Audible sound requires something more than the length to reproduce with good perceived quality. So, a VBR encoding can allocate very few bits to silent portions, and more bits to audible portions. The example of silence may seem silly, but there are lots of in-between cases that are harder to follow without serious math & audiology (I know the math, but not much of the audiology).

I haven’t studied the details of MP3 encoding, but I know that it was not designed with a careful consideration of what we know about aural perception, so the results at low bit rates, whether constant or variable, are noticeably weird to some listeners. At a high enough bit rate, MP3 usually sounds fine, and VBR is more likely to provide sufficient fidelity than CBR with the same file length.
Trapper
.
permalink   Mon, Nov 19, 2007 @ 10:39 AM
I totally agree. I’ve been using VBR encoding as my standard for several years now and, to my ears, it definately sounds better.
trap
 
.
permalink   duckett Mon, Nov 26, 2007 @ 11:46 PM
Not only does it sound as good, the little player on the page does NOT decide to freeze or drop out w/ VBR…
I cannot tell you how annoying it is to have to click play, then pause, waiting for it to buffer completely, in order to listen to someone’s track ALL the way through, w/o mood/time-killing pauses or drop-outs ;)
 
.
permalink   Surveillance_Party Sun, Apr 13, 2008 @ 6:40 AM
oh so I’m not the only one who gets that? Cause I get it alot.
 
.
permalink   duckett Sun, Apr 13, 2008 @ 7:54 AM
Actually, I haven’t gotten it since the latest version of ccMixter. Every once in a while, I might have to click the play button more than once to get the track to play, but that’s about it.