You can already do what you are aiming for, without the transcription part (removing fillers with word-based filters does not give great results compared to removing with voice-based filters).

You can use Descript, CleanVoice, and other tools to achieve exactly what you just said, in a few minutes, from just the original recording.