would be nice to have a way to connect another speech to text model, specifically AssemblyAI. Assembly is more accurate in comparison to Deepgram from my initial testing, and they work better with accents and languages other than english