Hey guys i hope you are doing well! Great work wi...
# ask-ai
l
Hey guys i hope you are doing well! Great work with LiveKit by the way! Do you have a recommendation which exact STT/transcription model with which exact settings works best for the german language? At the moment i am struggling pretty much. Single words are recognized wrong: as e.g. "Haarschnitt" is recognized as "Arschtritt" First and surnames are not recognized properly Phone numbers are also uncomplete many issues I ve tested "deepgram nova-3", "open-ai whisper-1", "open-ai gpt-4o-transcribe" Is there any way how i can add livekit openai stt support for primeline/whisper-large-v3-turbo-german (https://huggingface.co/primeline/whisper-large-v3-turbo-german) for example?
t
Thanks for reaching out, Denis. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
LiveKit offers support for several STT/transcription models that work great with the German language: • Deepgram: You’ll find models like nova-general, nova-2-general, nova-meeting, and enhanced-general, all supporting German (language code "de"). • Groq: They have models such as whisper-large-v3-turbo (which is the default), whisper-large-v3, and distil-whisper-large-v3-en. • Google Speech-to-Text: This one supports different German variants like de-AT (Austrian German), de-CH (Swiss German), and de-DE (Standard German). At the moment, there isn’t specific info about adding custom models like primeline/whisper-large-v3-turbo-german into LiveKit’s STT system. But if you have any more questions or want a hand with the models that are supported, just give me a shout—I’m here to help! Sources: plugins/openai/src/stt.ts | livekit-plugins/livekit-plugins-openai/livekit/plugins/openai/stt.py