Different Ways of Looking at Sound

Different Ways of Looking at Sound

I’m in the middle of ripping my cassette collection to MP3, and using Audacity to clean up the recordings, split them into individual tracks, and so forth.

One thing I’ve learned is that displaying a track as “Spectrum” rather than the default “Waveform” can be very useful in figuring out where one track ends and the next one starts, especially when the first track fades out and the next one fades in.

Here’s a perfect example I just ran across:

audacity-waveform.png

Here, the “S” label at the bottom was added by Audacity’s silence-finder tool, which I used as a first pass for detecting song breaks. At first, it looks fine: the song on the left fades out, and the “S” tag is 0.1 seconds before the drum beat that marks the beginning of the second song.

But when we look at this section as an FFT spectrum, here’s what it looks like:

audacity-spectrum.png

Here, it’s obvious that the silence-finder plugin missed the two quiet hi-hat notes that come just before the drum beat. On the waveform display, you can see the the hi-hat if you know it’s there, but it doesn’t really rise above the tape hiss enough to be noticeable, while on the spectrum display, it’s obvious.

Unfortunately, the spectrum display takes a while to update, but it’s oh-so-useful in closeups like this.