Optimal MP3 settings

[falcios]

How do I set the MP3 preferences in Audacity? I am new at this and I want to ensure the best audio quality.

Thanks in advance.

[allencmcbride]

For Audacity 1.2, at the bottom of the "file formats" tab under preferences, you can set the bit rate for MP3 export. I'm not sure if there are any more options than that. I know Audacity 1.3.7 has a lot more options than just bit rate. (If you try 1.3.7, you choose the export settings by selecting "export" from the file menu, entering any metadata, selecting "MP3 Files" from the "Format" drop-down menu, and then clicking "Options" next to it.) --Allen

[kozikowski]

What he said.

The encoder in 1.2 is nice, but you can do a much better job in 1.3.

Additional comments: Compressors are front loaded. The client always gets the show in real time and the quality they get depends entirely on the work you do and the time and care you take. Two-Pass takes longer and is better than One-Pass.

Do Not help out the MP3 encoder by compressing the show to fuzzy trash ahead of time. The encoder will preserve the fuzzy trash as much as possible. If you start with a top quality, perfect, clear show, the encoder will try to preserve that.

I believe the only way you can help out during production is to present a mono show instead of stereo. It will not save half the file size, but it will help and it will be significant.

One danger. Audacity wakes up from First Birthday in 32-bit floating, not 16-bit. 32-bit makes unstable MP3 files.

Koz

[allencmcbride]

Interesting; I didn't know about the 32-bit thing. So one should explicitly downsample to 16 bits before making an MP3? And I've never heard of one-pass vs two-pass... I see some promising Google hits, so I think I can figure out what the terms mean, but how do you choose these options in Audacity? I don't see them in the Options dialog. --Allen

[kozikowski]

Oops. My bad.

The two-pass thing is with M4A, not MP3.

I went down through all million pages of man lame and there is no option for dual pass.

However, I can document the others.

<<<So one should explicitly downsample to 16 bits before making an MP3?>>>

Yes. We had a posting where a voice performer complained bitterly that her customers couldn't listen to her MP3 demo clips. She sent one. I got it working and all I did was downsample to 16-bit and make a new MP3. "Unstable" MP3 in the sense that not everybody can play them.

Have you read the real 'man lame' (lame instruction manual)?

http://linux.die.net/man/1/lame

Lame can be used as a stand-alone program in linux-based systems, and if you do, the product can be fine tuned down to the last drumbeat. I'm particularly interested in one option about two thirds of the way through the document where you can select Generally Accepted Best Settings.

It's also a fact that people here on the board claim that under controlled circumstances, you can do better or as well with MP3 than you can with Apple's own AAC service based on the H.264 code. To which I rebut yes, you can work really hard in MP3, or just use the Apple presets. If you really pay attention to M4a (cousin to AAC) you can do amazing things. That's one of the technologies that creates the movie trailers on the Apple web site. I think that's how they're doing the HiDef movies on the TED web site.

Koz

[allencmcbride]

Thanks for the answer and the link. I actually installed the LAME application a couple weeks ago via macports, but I haven't really used it or studied it yet. It does look like there are some interesting options there. --Allen