AI Enhance Speech - how is Capcut so much better?

I’ve been using Pr and Au for a many years. Since I record on a boat I’m often contending with background noises like wind, engine, traffic etc, and getting clarity from my speech can be a long process until I reach something vaguely acceptable. The Enhance Speech, therefore, is a potential boon to speeding up my workflow. Except it isn’t. It’s awful. It introduces so many artefacts that speech ends up sounding distorted, tinny, adds echos etc. I have never been able to use this feature in Pr for a quick clean-up. Capcut’s equivalent Enhance Voice, on the other hand, absolutely nails it every time. It’s so good that I am currently running every piece of recording through it. Of course this adds to my workflow, but I just don’t understand how a premium product like Pr can’t come close to a cheaper alternative.

In both cases I am not touching any settings. Both default to 75% intensity. Does anyone have any tips on using Pr’s Enhance Speech, or is this yet another Adobe AI service that just doesn’t work? (I’m looking at you, Firefly.)

Many thanks.