REM is actually a family of tests, some are used for calibration or something, those with speach for help to adjust HAs. Ideally using calibrated recordings but can be done with real person, again ideally speaking through the device which helps to keep an eye on the output. That’s why recordings are better, every check it’s the same track, which is made to cover all vowels, in my case it looks like mix of english, french, hungarian, chinese and some other. I just remembered ‘pimyng the boss’ and ‘shorosh’.
Also, nerve damage could be if you hear distortion with both pure tone audiogram tones and in speech.
If your pure tones are tones without prrr brrr and similar addition (so lows are ummmm and not brrrr for example), then I’d be inclined to believe. It’s HAs settings.
Hope this helps a bit