me looking at the raw samples in audacity: "yes, that's clear as day; the round trip samples to the cassette and back still have two distinct waveforms, and all you need to do is count the zero-crossings because the signal is significantly louder than the noise!"
py-kcs:
6╠╚╚╫╚╚╚╚╚╚╚╚╚╚╚╚╚╚╚∞∞∞YÜY[█[K¢ô+ï½Kô+#cuse g directly.
22 80 ░░░ÉÜÖÉ╦╚╚╚╚╚╚3322
I guess I need to tinker with the decoder or something