In loud environment, you can hear speech better if it has the same volume all the time. Turning loud and quiet parts to the same volume is called dynamic range compression.

This article explains how to compress dynamic range of mp3 in Linux.

Dynamic range compression is good for listening speech, podcasts and lecture recordings. If you need to control volume level for hifi music, consider ReplayGain normalization instead.

Start with foo.mp3. Good candidates are lecture recordings with lots of changes of volume.

$ mpg123 -w foo.wav foo.mp3
$ sox foo.wav -t wav foo-companded.wav compand 0.003,.013.5 -80,-40,-60,-30,-40,-20,0,0 1
$ lame foo-companded.wav

And you end up with foo-companded.wav. It has it’s dynamic range compressed, so every moment is equally loud.

Sound quality is damaged by recoding and clipping. Some parts might sound like robotic voice.

Happy listening

Some hifi enthusiasts might say that I’ll spend eternity listening to compressed pop in Hell. Dynamic range compression takes life out of music, they say.

I feel that different mixing methods suit different situations. When you are listening to classical music, laying comfortably on your silent sofa and wearing headphones, you want to have the whole dynamic range. But when you are trying to listen to a speech recording in loud environment, you just want to hear the speaker.

You can also have your cake and eat it too: listen to your original lossless, full dynamic range FLAC at home, and have a completely compressed version on your mp3 player.

Jan Panteltje 2007: OT: preprocessing a mp3 file for playing in high noise environment

Wikipedia contributors: Dynamic range

SoX – Sound eXchange – HomePage