Beyond the hiss

The tape on a compact cassette imposes all kinds of noise and distortion on a recorded signal. Much of it is so quiet that it's practically inaudible, but with some signal processing we can zoom into some fascinating details.

[Image: A spectrogram, 0 to 22 kHz, of a 5-second test signal recorded on a cassette tape and played back.]

(There's a high-res version of the enclosed infographic at Dropbox.)

I generated a test tone that sweeps from 50 Hz up to 21 kHz in 5 seconds, then recorded it onto a cassette. I played the tone back and plotted it above, in turquoise. I also monitored the "dry" signal passing through the circuitry and plotted it in red, to see what was actually caused by the tape.

There's a prominent banding that seems to follow the rising frequency of the tone. The pattern, with its mirrored sidebands, looks very similar to a frequency-modulated carrier, and that's effectively what it is. The tone frequency is being modulated by rapid fluctuations in tape speed during recording and playback, also known as flutter.

Another persistent signal is visible in the leftmost part of the picture, or the low part of the spectrum. It's the humming sound of the power grid, 50 Hz in this part of the world, leaking to the tape because of an unwanted ground loop somewhere in the circuit. Ultimately, it's the sound of the steam turbine rotor turning at the nearest power station. The hum has several harmonics at integer multiple frequencies.

Some harmonic multiples of the test tone are also visible. These are caused by gross distortion of the original waveform, most probably by magnetic saturation (clipping) of the tape media, or they may be crossover distortion from an amplifier stage. The harmonics are repeated in the end of the sample as a zig-zag pattern, which is probably a digital aliasing artifact.

The noise floor on the cassette is set by what is commonly known as tape hiss. It is caused by the finite size of the magnetic particles that the audio is recorded on. This is similar to the grainy pattern on an enlarged film photograph, or pixels in a digital one. A faster-running tape will have more hiss, but at the same time can record information at a greater bandwidth.


  1. Thank you so much for sharing this. Very useful!

  2. Just discovered this blog entry and poster. much of what you are terming "flutter" is actually an artifact of the high frequency *bias* mixed with the signal to the recording head??? Looks too much like subsampled higher frequencies to me, and that much real flutter/variation would seem to me to be easily audible. But I'm not claiming to be an expert on this.

    Take a look at the Wikipedia article on Tape Bias. Also, I assume the cassette drive you were using didn't have any kind of noise reduction (e.g. Dolby B), which would introduce its own set of artifacts.

  3. This is wonderful. Very cool. Thank you!


The comments section is pre-moderated; it will take some time for the comment to show up.

You might want to check out the FAQ first.