The distinct midrange tone in this recording varies from about 870 to
950 Hz. Spectrogram shows it as a series of scallops. There's a wide
curve as it drops to 870 Hz, then starts rising again. When it reaches
950 Hz, the "creak" happens, and the tone frequency starts dropping.
I think this was a cheap cassette machine, and the cassette was
jamming. I assume that playback speed was constant when this transfer
was done. If the speed was varying during recording, high tape speed
during recording = lowest tone frequency on playback, and vice-versa.
When the tone is around 870 Hz, recording speed was normal. As the tone
rises in frequency, the tape was slowing down during recording. I think
the "creak" is either mechanical noise from the cassette, picked up by a
microphone built into the recorder, or electrical noise generated by the
capstan motor governor when it freaks out because it can't rotate the
capstan at the correct speed.
Whatever the cause, it's clearly recorded on the tape. The spectrum of
the "creak" extends up to about 10 kHz. The voice sounds have a similar
upper limit. If this noise was generated during playback, it would
probably extend all the way up to 20 kHz. But I can't imagine how it
could have been generated during playback.
-- John Chester