|dreams

How to encode your MP3 podcast files using linux

I follow these recommendations from the podcaster mailing list.

44.1 kHz sample rate

“Sample rate first: you should use a sample rate of 44.1 kHz if music fidelity matters to you.”

“Oh, and a final note: never use a sample rate that isn’t a “multiple of 11″ rate. Flash-based Web players have a glitch in them so that any other rate (e.g. 48 kHz or 24 kHz) will play like Alvin & the Chipmunks. Unless you’re Jonathan Coulton producing “A Podsafe Christmas Song,” you don’t want this.”

128 bps bit rate

“In my opinion, the three most reasonable bit rates to choose from are 64 kbps (for a spoken word podcast), 96 kbps (for mixed) and 128 kbps (for stereo music). Anything less than 64 kbps and the compression loss is really noticeable. Higher than 128 kbps is just wasting bandwidth. Again, try all three with your content, and use whichever one feels right to you.”

No VBR

“Oh, and you didn’t ask, but you should always use joint stereo, and never use variable bit rate (VBR). Joint stereo gives the best quality and compatibility regardless of whether your signal is mono or stereo; and VBR is glitchy on some MP3 players.”

What this boils down to if you’re using lame

If your input wav file is already 44.1 kHz and stereo you run lame without options since it will default to 44.1 kHz, joint stereo, 128 bps and no VBR. If you’re unsure what kind of wav file you have inspect it with the file command

file input.wav

If everything looks fine then just do:

lame input.wav

If your sample rate is wrong you’ll have to resample

lame --resample 44 input.wav

If your input file is mono lame will give you mono 64 kbps output which is not what you want.

lame -m j -b 128 input.wav

You’re done.