Subject:MP3 Compression Quality

Posted by: BarryBond

Date:7/6/2004 1:49:46 PM

I don't know if this is asked before, sorry if it has.

I'm doing an audio demonstration at work and I was just curious as to what exactly happens when you adjust the compression quality setting when compressing to MP3. I understand that it takes longer with a better setting meaning that you get better compression, but can someone please tell me exactly what this does. Does it calculate the ailising better, make the bits more accurate?

Thanks,
-b

Subject:RE: MP3 Compression Quality

Reply by: Geoff_Wood

Date:7/6/2004 4:06:54 PM

Throws away slightly less wanted information. Less phasy artifacts and missing detail. I consider 160kbps to be comparable to cassette (but without the hiss).

geoff

Subject:RE: MP3 Compression Quality

Reply by: BarryBond

Date:7/7/2004 12:27:06 PM

Actually, I was talking about the quality slider rather than the compression setting. I mean, I realize that with the slider pushed all the way to the right setting that it takes longer to encode, which means less encoding errors, which equals a better quality file, but I was hoping someone could explain exactly how this works from a program perspective. What goes on inside the program to make it better quality. I know this is kind of difficult to explain. Sorry if I'm being too vague.

Thanks for your response.

-b

Subject:RE: MP3 Compression Quality

Reply by: rraud

Date:7/7/2004 3:49:04 PM

If this is any help:
A 64kbs MONO mp3 file is equivelent in quality to a 128kbs STEREO mp3 file. ... Unlike a .wav file which in stereo; is x2 the mono file size.

There is also different algorithms. I currently favor the r3-WinLAME which can batch convert as well as offer adjustable hi-pass/lo-pass filter points among the many other normal and not-so-normal parameters.. through some must be changed in a .xlm code file.

Subject:RE: MP3 Compression Quality

Reply by: Rednroll

Date:7/7/2004 6:44:10 PM

Actually, the question has to do with the Variable bit rate quality adjustment slider, which has a quality setting. This slider is only available during VBR encoding. I have no idea, what goes on behind the scenes in the quality calculations. I have always thought of it, as to being similar to the winzip compression techniques. In Winzip you can choose things like fast, but lower compression, or slow most optimized compression. I'm sure it's a similar kind of thing, but I have no idea what that is. Good question, I'ld be curious in knowing also.

Subject:RE: MP3 Compression Quality

Reply by: Chaim

Date:7/8/2004 5:08:03 PM

I don't know the inner workings of the program, but I do know that MPEG (mp3) compression is extremely asymetrical, i.e. it takes a tremendous amount of work (read "compute cycles") to compress, and very little work to decompress, which is why little gadgets like iPods can easily decompress and play back mp3 files. A slider in the compression gadget which decreases the time to complete the job, would almost certainly decrease the quality of the finished product.

Subject:RE: MP3 Compression Quality

Reply by: jorgensen

Date:7/19/2004 1:01:52 AM

For SoundForge VBR seams to work like following.

A VBR of 50 sets the maximum bitrate to 128kB, and is adjusted downwards when possible. A VBR of 100 sets the maximum to 256kB (or something like that).

You can use WinAmp to see the actual bitrate.

Subject:RE: MP3 Compression Quality

Reply by: BarryBond

Date:7/20/2004 11:17:18 AM

Actually, this question is in reference to the quality slider when doing CBR encoding. I understand the quality slider in VBR, but there's also one when you want the bit rate constant. I think I sort of get the idea of it. Obviously longer encoding times = better quality and better encoding. I'm starting to confuse myself. I need a cup of coffee.

-b

Subject:RE: MP3 Compression Quality

Reply by: jorgensen

Date:7/20/2004 10:46:20 PM

The post was in response to Rednroll, who as I, have been confused what the VBR control settings precisely meant.

The quality fader indicates, how much processing time you want to use on the encoding. The more processor time, the better quality, but I doubt you can see/hear a difference for standard encoding.

Subject:RE: MP3 Compression Quality

Reply by: Rednroll

Date:7/21/2004 6:02:03 AM

"Actually, this question is in reference to the quality slider when doing CBR encoding."

CBR? When I select a CBR setting my quality control gets greyed out. Only when I select VBR does that quality control become adjustable, so that's why I assumed the question was regarding VBR.

Now I'm totally confused?

Thanksa for the explaination on the VBR Jorgensen, this makes sence to me now.

Subject:RE: MP3 Compression Quality

Reply by: Chienworks

Date:7/21/2004 6:36:14 AM

Red, there are two Quality sliders. In the MP3 custom template screen the first control is Bit rate and the third control is VBR Quality. Between those two is a "Quality" slider with three positions: Fastest encode, Medium, and Highest quality. This slider is active when doing either VBR or CBR encoding.

Presumably the faster encoding algorithm is faster because it takes shortcuts that don't do as good a job and the slower encoding algorithm takes longer because it is doing a better job. I've found that i notice a big improvement in medium over fastest without taking too much more time, but highest quality doesn't sound that much better than medium and takes an awful lot longer. I've been leaving it set at medium most of the time.

Subject:RE: MP3 Compression Quality

Reply by: farss

Date:7/24/2004 6:46:46 AM

Not 100% certain but according to the makers website the encoder has a number of loops in whcih it tries to resolve the best fit between the amount of detail in the audio it has to encode and the available bitbudget. I'd expect the slder controls the number of iterations the loop makes as it tries for the best fit.
You see much the same with mpeg-2 encoding. At higher bitrates the 'quality' settings don't amount to much difference, as you push the bitrate down giving the encoder as much time as possible gives it a better chance at getting the optimal result.

Bob.