https://livekit.io logo
Join Slack
Powered by
```cant get the user transcription when using gemi...
# ask-ai
s
Copy code
cant get the user transcription when using gemini live api. how to fix??
t
Thanks for reaching out, SajithlalDev. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
It seems like you are experiencing issues with user transcription when using the Gemini Live API. Here are some steps to help you troubleshoot: • Ensure that the 
user_transcription
capability is enabled in the Gemini Realtime model. This can be configured through the 
input_audio_transcription
parameter when initializing the RealtimeModel. • Verify that the 
enable_user_audio_transcription
option is set correctly. If enabled, a 
TranscriberSession
should be created to handle the transcription. • Check if the audio input is being processed correctly and that there are no issues with the audio format or quality. If the issue persists, please provide more details about your setup and any error messages you might be encountering, so we can assist you further. Sources: Tree View: | dse-workbench | Tree View: | dse-workbench
s
i am using v1
t
Thank you for clarifying that you are using version 1 of the Gemini Live API. In this version, if the 
user_transcription
capability is enabled, the regular Speech-to-Text (STT) transcription is skipped. This is by design, as the Gemini transcription process takes over. To resolve the issue, please ensure the following: • Verify that the 
user_transcription
capability is properly enabled and configured in your setup. • Check if the Gemini transcription process is correctly implemented and receiving the audio input as expected. • Ensure that there are no conflicts or errors in the configuration of the transcription settings. If you continue to face issues, please provide additional details about your configuration and any error messages you are encountering, so we can assist you further. Sources: Tree View: | dse-workbench
Open in Slack
PreviousNext
Powered by