https://livekit.io logo
Join Slack
Powered by
Google STT not working with Telugu
# ask-ai
g
Google STT not working with Telugu
t
Thanks for reaching out, Pavan. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
I'm glad to let you know that the Google STT integration in LiveKit does support Telugu (te-IN), and the underlying Google Cloud Speech-to-Text service supports it as well. To help you troubleshoot this, here are a few things to check: • Make sure you're using the correct language code 
te-IN
when initializing the STT instance. • Try using the 
chirp
model, which can often improve transcription accuracy for regional languages. • Double-check that your Google Cloud project has an active billing account and that the Speech-to-Text API is enabled. • Confirm that your Google Cloud credentials are set up correctly. If you’re still experiencing issues, a helpful next step is to test Telugu recognition directly through the Google Cloud Speech-to-Text API demo. This can help pinpoint whether the challenge is with your Google Cloud setup or the service itself. Please feel free to reach out if you need any more guidance—I’m here to help! Sources: livekit-plugins/livekit-plugins-sarvam/README.md | Google integration guide | LiveKit Docs | Google Cloud STT integration guide | LiveKit Docs
g
we are using te-IN tried with crisp too and have an active account
t
Since you're using 
te-IN
with the 
chirp
model and have an active account, let's try a few more steps together to get this sorted out: • First, please double-check that your Google Cloud credentials are set up correctly—whether that's through 
credentials_info
, 
credentials_file
, or Application Default Credentials. • Next, make sure your audio input matches the required sample rate of 16kHz and is clear enough for accurate processing. • It might also help to enable verbose logging in your application. This way, you can see if the language code is properly sent to Google's API and catch any specific errors that pop up. If you’re still running into issues after trying these steps, just let me know. I’m here to help, and we can dig deeper to find a solution together! Sources: livekit-plugins/livekit-plugins-baseten/livekit/plugins/baseten/stt.py | Google integration guide | LiveKit Docs | Google Cloud STT integration guide | LiveKit Docs
g
can i use gpt-realtiem just of stt? and use different llm and tts? in livekit
t
Absolutely, you can definitely use GPT-realtime just for speech-to-text (STT) while bringing in a different large language model (LLM) and text-to-speech (TTS) provider within LiveKit. Here’s how you can do it: • Set the 
modalities
parameter to 
["text"]
in the GPT-realtime configuration so it focuses on outputting text only. • Bring in a standalone TTS plugin like ElevenLabs or Cartesia to take care of speech synthesis. • If you’d like, you can also send the transcribed text through your preferred LLM before passing it on to the TTS service. This way, you have the flexibility to mix and match components for a pipeline that perfectly fits your needs. If you’d like, I’m here to help with any specific configurations or examples you might need! Sources: Realtime model integrations | LiveKit Docs | OpenAI Realtime API integration guide | LiveKit Docs | Azure OpenAI Realtime API and LiveKit | LiveKit Docs | Gemini Live API integration guide | LiveKit Docs | examples/src/cartersia_tts.ts
g
can you give me the code with realtime stt and gemini llm and tts
plz give
Open in Slack
PreviousNext
Powered by