This combo has almost unbeatable accuracy and it rejects noises in the background really well. It can even reject people talking in the background.
The only better thing I've seen is Ursa model from Speechmatics. Not open weights unfortunately.