Just spotted this nifty cross-platform (if your platform isn’t linux…) app via makezine and had a go.
Its a simple drag and drop application which analyses your recording and brings the speech throughout it up to the same level. I tried it out on a recording where I had three informants talking, and one of them whispered his way through it. Here’s a little sample of this session. The third person speaking is the one who whispered his way through the session. Here’s the original for comparison. Normally I’d normalise the file, so the levels are probably a little low in the original, but also normalising looks at the whole file, not individual sections of the file, so if someone was yelling later in the recording, the softer sections would basically stay the same.
Obviously, the background noise and hiss goes up with each amplified section, but you can’t complain about that! It’s slightly more sophisticated than normalising your files, and just as easy, so this is another tool I’ll be adding to my arsenal. Of course, it goes with out saying that you should use this to create temporary working files for transcription, but never delete your originals.
See also: Get rid of those cicadas!