Fish Audio has released Live 3, a full-duplex voice model built for real-time voice agents. The Fish Audio Live 3 full-duplex voice model keeps listening while it speaks. That lets a caller pause, say “mm-hmm,” or change their mind without breaking the call. For developers, the pitch is simple. Turn-taking moves out of hand-tuned pipeline rules and into the speech model.
TL;DR
- Product: A full-duplex voice model built for real-time voice agents
- Runs on: Hosted Fish Agent Platform and Fish Audio API; on-premises for enterprises.
- Performance: Fish Audio shows behavior demos, but publishes no benchmark scores.
- Best: Can decide to respond in as little as 186 ms for short reactions like “mm-hmm.”
What is Fish Audio Live 3?
Live 3 is a full-duplex voice model from Fish Audio that listens and speaks at the same time.
The core argument is that speed and timing are different problems. An agent can reply fast and still speak at the wrong moment. A pause after “What time do you open?” usually ends a turn. A pause in the middle of a phone number does not.
How does Live 3 handle turn-taking?
Most voice agents split recognition, reasoning, speech generation and turn-taking across separate pipeline stages. Fish Audio team states that Live 3 moves conversational timing into the speech model. It was trained on real human conversations to learn when to acknowledge, wait, respond or stop.
They framed the old approach as piling on “more rules and more thresholds.” Fish Audio instead wants the model to learn timing from data. A separate reasoning layer still holds the agent’s instructions and tools.
The Fish Audio team walks through 3 behaviors:
- Acknowledging without taking over. A caller reads “415… 555… 0132.” Live 3 may say “okay, 415” between chunks, then wait.
- Telling a backchannel from an interruption. A caller saying “yeah” lets the agent continue. “Wait, actually, I just need Thursday” makes it stop.
- Waiting through corrections. For “two pizzas… no, three,” Live 3 waits instead of logging the wrong order.
How fast is Live 3?
Fish Audio states Live 3 can decide to respond in as little as 186 ms. That figure covers short reactions such as “mm-hmm,” “okay” and quick confirmations. The full answer is prepared in parallel.
This is a floor for small reactions, not time-to-first-audio for a complete reply. Fish Audio has not published a latency distribution. Third-party tracker OrcaRouter also reports no Artificial Analysis score for Live 3 as of November 11.
