This may not help any as I cannot remember the name of the program, but I swear that about 10 years ago I had a CW learning program that would let you insert white noise as you listened. I believe you could choose how much noise you heard, to simulate different conditions. I'll look around and let you know if I figure out what it was.
Here is a set of audio files with -10 dB to +20 dB SNR @ 2kHz https://www.dropbox.com/sh/2j9chxsizxjeqgp/Gj97asYzpW
that I just created for testing a new Bayesian Morse decoder. The files contain 50 WPM random text (see random.txt on the shared folder) - 200 words of 5 letter/number each.
Let me know if these are useful.