Training a computer to listen and decode Morse code (2019)
ag1le.blogspot.com
ag1le.blogspot.com
IIRC, the biggest problem was working out the speed of the sending. That took some trial and error, but eventually gave some text output to the screen.
A similar project worked on RTTY signals, that worked better than the morse project because RTTY signals are sent mechanically and don't have the 'human variability' factor.
Going back to the article, humans do better on noisy signals because they can naturally distinguish close audio tones. The electronic way would be to pre-process the noisy audio signal with fourier analysis filtering the noise from the signal before converting it to binary levels.