Can I use the iOS Speech Framework instead of STT?
# ask-aib
bulky-wall-61623
09/13/2025, 10:36 AMCan I use the iOS Speech Framework instead of STT?
t
tall-airline-16756
09/13/2025, 10:36 AMThanks for reaching out, RikuKomiya. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
tall-airline-16756
09/13/2025, 10:36 AMThank you for reaching out! I wanted to share that the iOS Speech Framework isn’t designed to be a direct replacement for the Speech-to-Text (STT) systems in LiveKit. LiveKit currently supports third-party STT providers like AssemblyAI, Speechmatics, Deepgram, and OpenAI Whisper, but it doesn’t integrate with the iOS Speech Framework just yet.
If you’re interested in using device-native speech recognition, you would need to set that up manually outside of LiveKit’s STT plugin system. Please feel free to ask if you’d like me to walk you through this or if you have any other questions—I’m here to help!
Sources: agents/src/stt/stt.ts | AssemblyAI integration guide | LiveKit Docs | Speechmatics integration guide | LiveKit Docs | STT | LiveKit Agents | livekit-plugins/livekit-plugins-spitch/livekit/plugins/spitch/stt.py
b
bulky-wall-61623
09/13/2025, 11:31 AMCan I use whisperx?
t
tall-airline-16756
09/13/2025, 11:31 AMI understand you're looking to use WhisperX with LiveKit, and while LiveKit doesn’t support WhisperX out of the box, there are definitely ways to make it work through a custom setup. Here are a couple of friendly options you might consider:
• Deploy WhisperX on Baseten: Baseten is great for hosting custom machine learning models like WhisperX. You can easily connect it to LiveKit by using the Baseten STT plugin, which serves the model over WebSocket at this endpoint:
wss://<your-model-id>.<http://api.baseten.co/v1/websocket|api.baseten.co/v1/websocket>