Full-duplex audio
Accepts streaming audio input and returns incremental spoken output.
AI Models
Qwen Omni
Alibaba Cloud's end-to-end Qwen Omni realtime speech model for live phone agents. Stream call audio in and play spoken replies back in the same session.
Speech-to-speechRealtimeAlibaba CloudMultilingualFunction tools
Where it fits
Qwen Omni in your voice stack.
- TelephonyPhone networkInbound call
- AgentDuetCall + audioAnswer, stream
- Speech-to-speechQwen OmniListens, call functions, replies
- ToolsActionsCRM, calendar, APIs
Overview
What it contributes to the stack.
Qwen Omni Realtime is Alibaba Cloud's end-to-end speech model for live conversations. It supports continuous spoken dialogue, turn detection modes, and function tools.
With AgentDuet, the phone or WhatsApp connection stays on AgentDuet while your application bridges call audio to Model Studio and plays spoken replies back.
For setup, credentials, and API details, use the Documentation link above.
Capabilities
The technical characteristics that matter when deciding whether this provider belongs in your application.
Turn detection modes
Supports server voice activity detection, semantic turn detection, and push-to-talk style control.
Function tools
Requests approved application functions with structured arguments.
Configurable voices
Selects system or cloned voices for generated speech responses.
Common use cases
Practical scenarios where the provider's role is clear and the surrounding systems remain under application control.
Voice assistants
Run continuous spoken conversations with low-latency replies.
Contact center automation
Answer routine requests and resolve complex inquiries with collected context.
Multilingual support
Serve callers with Qwen Omni realtime speech models suited to language coverage needs.
Related integrations
Compare nearby choices.
These providers occupy a similar role in the stack. Compare their strengths, operating model, and surrounding services.
Browse every integration
Build a voice agent.
AgentDuet handles the phone and messaging boundary. Your model runs the conversation.