What’s New
-
Configure 16 text normalization processors in FormatPlan: You can now control how text is transcribed and spoken for currency, dates, etc. by setting the
formattersEnabledarray inAssistant.voice.chunkPlan.formatPlan(not specifyingformattersEnableddefaults to all formatters being enabled). See all available formatters in the FormatPlan.formattersEnabled reference. -
Deepgram Keyterm Prompting: The
keytermarray in DeepgramTranscriber implements Deepgram’s Keyterm Prompting technology, boosting recall for domain-specific terminology. Compared to the existingkeywordsfield:
You should reserve keyterm for compliance-sensitive terms like medical codes while using keywords for proper nouns / brand names.
-
Subscription usage tracking improvements: The
minutesUsedNextResetAttimestamp now appears in all subscription tiers (not just enterprise), exposed atsubscription.minutesUsedNextResetAtfor predictable billing cycle integration. Combine with existingminutesUsedandminutesIncludedmetrics to build custom usage dashboards, regardless of subscription tier. -
Neuphonic voice synthesis: You can now configure Neuphonic as a voice provider with
Assistant.voice[provider="neuphonic"]. Handle appropriate errors withpipeline-error-neuphonic-voice-failed. Test latency thresholds as Neuphonic requires 200ms additional processing time compared to ElevenLabs.

- Support for pre-transfer announcements in ClientInboundMessageTransfer: The
contentfield inClientInboundMessageTransfernow supports pre-transfer announcements (“Connecting you to billing…”) before SIP/number routing. Implement via WebSocket messages using type: “transfer” with destination object.