Quick config
tools_config.transcriber block. See Create Agent API for the full schema.
To use your own Deepgram account, connect it at platform.bolna.ai/auth/deepgram.
Supported models
Recommendation: Use
nova-3 for English agents. Use nova-2 with language set explicitly for non-English.
Key settings
Endpointing guidance
- 250ms — default; works well for fluent English speakers
- 350–500ms — better for non-native speakers, elderly callers, or callers who pause mid-sentence
- 100–150ms — fast-paced scripts; reduces dead air but may interrupt on pauses
Multilingual config
Setlanguage explicitly. Auto-detect adds 100–200ms latency and occasionally misclassifies.
FAQ
Should I use nova-3 or nova-2?
Should I use nova-3 or nova-2?
Use
nova-3 for English — it’s faster and more accurate. Use nova-2 when you need a non-English language or a domain-specific variant (nova-2-phonecall, nova-2-medical, etc.).Why is my agent interrupting callers mid-sentence?
Why is my agent interrupting callers mid-sentence?
Your
endpointing value is too low. Increase it to 350–500ms. If callers have a strong accent or speak with filler words (“um”, “uh”), a higher endpointing value gives them more time to complete their thought.What encoding and sampling_rate should I use?
What encoding and sampling_rate should I use?
Match the telephony provider’s output: Plivo and Exotel use
linear16 at 16000 Hz. Twilio uses mulaw at 8000 Hz. If you use the wrong values, transcription accuracy degrades.Does Deepgram work with inbound calls?
Does Deepgram work with inbound calls?
Yes — the same transcriber config applies to both outbound and inbound calls. There’s no separate configuration.
Related
- Audio Tab — configure transcriber in the dashboard
- Latency — how transcription affects end-to-end response time
- Deepgram Flux — the Flux model variant
- Sarvam transcriber — Indian languages
- Azure transcriber — enterprise multilingual option