If your multilingual AI Employee (e.g., Russian/English, Spanish/English) experiences decreased transcription quality or difficulty understanding client speech, you can improve accuracy by adjusting the speech recognition model in the connector settings.
In the Builder, navigate to the Integrations page.
Click the three dots icon next to the
newo_voice_connector.Select Edit Settings.
Update the below recommended settings:
Voice To Voice: User STT Provider:
googleGoogle STT Model:
chirp_2Google STT Language codes:
auto
Notes
If you use only one language, specify the corresponding language code (BCP-47) from this table instead of auto.
The selected model chirp_2 must support the chosen language. If it doesn’t, select another model in the Google STT Model field.
The chirp_2 model shows the best results compared to others and is recommended as the primary choice.