not just constant level, constant frequency. The Mains will be constant enough, but the tape may not be. The apparent frequency (and hence phase) will drift up or down with the stretched parts of the tape.

"noise reduction" in Audacity has not worked for me that well. What any subtractive process will do is increase the hiss because it is not coherent, by its nature it is random. Adding random to random is random X2. So you end up cutting the high end anyway.

Just a thought, but if you are patient enough you could try notch filters in all the silences. Works best on speech and acapella. Music tends to be pretty much "on all the time" so the effect would be very noticeable even if you found any relative silences. There is a lot of "perception" in audio, hum in the quiet sections dominates, but is "relative" in the louder passages. The wheeze above can work, I do it if it improves the overall result.

Best of luck.