Some simple ABR vs VBR question Options

[knuck]

I've been reading about ABR on the internet and I've read lots of "VBR will be slightly better at the same bitrate".
Firstly, I'm not an audiophile. I can't hear anything past 160kbps. Secondly, I like to optimize my space usage. I could use VBR, but I've seen VBRs using more than 160kbps a lot of the time in most songs, and that'd be a waste of space.
So, I'm wondering which ABR setting should I use so that I have a maximum bitrate of a 160kbps CBR file?
Using LAME by the way.

And yeah I'm a noob at audio stuff.

[rpp3po]

I've been reading about ABR on the internet and I've read lots of "VBR will be slightly better at the same bitrate".
Firstly, I'm not an audiophile. I can't hear anything past 160kbps. Secondly, I like to optimize my space usage. I could use VBR, but I've seen VBRs using more than 160kbps a lot of the time in most songs, and that'd be a waste of space.
So, I'm wondering which ABR setting should I use so that I have a maximum bitrate of a 160kbps CBR file?
Using LAME by the way.

And yeah I'm a noob at audio stuff.

VBR delivers more quality per byte than ABR. If you find a specific VBR setting too high on average over your whole collection lower the Q value. When a VBR encoder allocates surprisingly high bitrates to single tracks it does so for a reason (hard to encode without artifacts). Let it do its job. On average over your whole collection the bitrate will still be lower than ABR at comparable quality.

This post has been edited by rpp3po: Feb 26 2009, 04:11

[lesswire]

Hi, you could download a software called foobar2000 to do some ABX tests, just make sure to enable the ABX function when you install it. Convert some of your favorite tunes (from lossless sources) at different configurations and use the ABX test utility to find out which configuration suites you better. When you find that a particular configuration yields small file sizes and transparency, you will have found your ideal configuration. VBR files tend to fluctuate more than ABR but they have an overall better quality. Hope this helps.

[halb27]

With a target bitrate of 160 kbps I'd use Lame 3.98.2 VBR.
What counts is average bitrate for your typical music, and in order to find out I'd build a small but representative track set and encode it.

I guesss something like -V4 is an adequate setting, and you are not restricted to integer values for the -V setting. Maybe -V4.5 or -V3.5 matches your needs better than -V4.

In case you are not too sensitive towards temporal resolution as well as extremely high frequencies in music you may want to consider using the unusual setting Lame 3.98.2 -V3 --lowpass 15.2 which yields an average bitrate of 150 kbps for my typical music test set consisting of pop music of various kind and age. The very high frequencies thus are restricted to ~15.2 kHz (which equals pretty much the HF behavior of good FM radio), and music is downsampled to 32 kHz. 32 kHz sample frequency makes encoding more efficient allowing for higher quality settings for the same bitrate. Eventual artefacts in music of the tonal kind (to me the ugliest kind of errors) are lowered significantly, on the down side temporal resolution gets worse.
Even -V4.5 --lowpass 15.2 (yielding 130 kbps for my test set) gives very good results for my ears. You can try for yourself in case this sounds attractive to you.

This post has been edited by halb27: Feb 26 2009, 11:15

[Remedial Sound]

To the OP, I echo what others have said here in that you can go the preferred quality route in VBR and still have confidence that most of your encodings will be within a certain bitrate. I've been using Lame -V4 for a couple of years now and can say that on average the bitrate is around 160 kbps. Granted, you might average 180 over one album and 140 over another, but at the end of the day by going with one of the presets you're ensuring that all of your encoding are generally the same with respect to quality.

[knuck]

Thanks guys for the replies!

VBR delivers more quality per byte than ABR. If you find a specific VBR setting too high on average over your whole collection lower the Q value. When a VBR encoder allocates surprisingly high bitrates to single tracks it does so for a reason (hard to encode without artifacts). Let it do its job. On average over your whole collection the bitrate will still be lower than ABR at comparable quality.

This explanation makes lots of sense (hard to encode without artifacts) and I have decided to use VBR over ABR. =)

[kornchild2002]

I thought that people weren't advised to adjust the different -q values in Lame. In other words, if you want to lower the overall average bitrate, simply decrease the -V value. I think that -V 4 will produce adequate results for your needs as, for the most part, music comes out with an overall average bitrate at around 150kbps. I tested -V 4 (a while back) on a few files of mine and the bitrate it produced was a little above 160kbps. These were with songs in the metal genre (Lamb Of God, Static-X, and HELLYEAH) and they tend to make encoders use a higher bitrate when encoding (whether I encode them with Lame mp3, Nero AAC, or iTunes AAC).

[MichaelW]

@kornchild: I think repp3po was using "Q value" in a generic way, that would translate into LAME's V settings, rather than the q switch. There, glad to have cleared that up for you.

To OP: if you look at the recent ~128 kbps test and discussion, you'll see explanations of why all the test samples have bitrates above 128; the average bitrate was set over a corpus of music, but test samples tend to be difficult to encode, and there you can see VBR working.

[rpp3po]

@kornchild: I think repp3po was using "Q value" in a generic way, that would translate into LAME's V settings, rather than the q switch. There, glad to have cleared that up for you.

To OP: if you look at the recent ~128 kbps test and discussion, you'll see explanations of why all the test samples have bitrates above 128; the average bitrate was set over a corpus of music, but test samples tend to be difficult to encode, and there you can see VBR working.

MichaelW is correct. It's Q for AAC (Quicktime, Nero) and V for Lame.

[timcupery]

to the OP: if you want every file to be around 160 kbps, then use ABR.
but a better way to proceed is pick a VBR quality-level that you're happy with, and go with that. For example, if you use V4 (which will average around 160 kbps, depending on the style of music), then you'll have some songs that encode at around 120 kbps, and others that encode close to 200 kbps. This is okay, and you can feel better that you're getting consistent musical quality.
Whereas, if you used ABR 160 to encode some pan flute music, you'd be way, way overkill, whereas ABR 160 on complex heavy metal might have some audible artifacts even for you.

Based on what you've said about your hearing, you'd probably be fine with Lame V5, but could go with V4 just to play things safe.
If you want to go through the effort, do some ABX tests (and get ahold of some problem samples that are difficult to encode without artifacts if you want to do this well).