v2.7.0VoiceAugust 24, 2026
Voice AI Engine 2.0: Sub-500ms Conversational Telephony
Voice AI Engine 2.0 delivers natural, conversational phone interactions with ultra-low latency audio streaming and full-duplex speech recognition.
Key Release Highlights
- Token-streaming audio pipeline reducing telephony turn-taking delay to 480ms
- Intelligent barge-in detection allowing callers to interrupt seamlessly
- Support for 12 regional languages and contextual accent adaptation
- Real-time automated call summarization and sentiment score webhooks
Our new speech processing pipeline processes audio chunks in real-time over bi-directional WebSockets, eliminating awkward pauses during phone conversations.
### Key Improvements:
- **Natural Barge-In**: When callers speak while the AI is responding, speech synthesis pauses within 60ms to listen.
- **Accurate Domain Terminology**: Fine-tuned phoneme vocabulary for BFSI, healthcare, and e-commerce terms.
- **Live Transfer to Human Desk**: One-click SIP trunk transfer preserving complete conversation context.
Want to start using this feature?Read Feature Documentation