The last few days, I’ve been working on recovering some recordings of a ten year old interview with Buck O’Neil, of the Kansas City Monarchs of the old Negro American League, during segregation. He lead the effort to establish the Negro Leagues Baseball Museum, and has several other firsts; Ken Burns used him quite a bit in his documentary on baseball.

The recordings were made on a pocket microcassette recorder, ten years ago. I don’t know if you’ve ever seen one of these things, but they were one of the smallest tape factors ever made, moved really slowly, and did not age well.

Even optimally, they could record about 50hz to 4000hz – less than a fifth the range of modern equipment – and are prone to noise, hiss, and other problems, all of which this recording had. Plus, the recording was made handheld, via a PA, which itself had feedback issues causing ringing on emphasised words, all at highly variable signal levels.

So it’s been quite a learning experience. Knowing much more about this from theory than practice, I was glad to get a chance to try out some of these tricks.

I used a repetitive-noise pattern-matching filter to etch out the worst of the repeating noise profile. Then a low pass filter to get rid of everything above 4000hz and a high-pass filter to get rid of everything below 50hz; since those are beyond the capability of the format, anything in those regions was playback harmonics or noise.

Then a collection of notch filters to get rid of primary feedback tone and two common harmonics of it, variable-band equalisation filters to duck out what I could of the recorder’s noise and some in-room sound, and some very careful but high-ratio compression to bring it all to a vaguely consistent level. In some areas, this exposed or exaggerated sibilants, which I filtered out. Then I threw on (in some places) a tiny touch of reverb to hide the worst of the tape-jitter distortion and make it all understandable, followed by a final equalisation round to try to throw a little life back into it, and another round of compression.

The result isn’t something I’d call good – you can’t get there from here – but it’s reasonably clear, has a lot less noise, and listenable, whereas the originals range between better than you’d expect and inaudible, with a deeply fatiguing feedback ring.

I may be able to link to the result, eventually. If you get a chance to hear a recording of Buck O’Neil at one of these sorts of events, you should; he’s got a million stories from segregation and baseball, and they’re all interesting.