Nemotron VoiceChat 11B
A full-duplex voice AI model released by NVIDIA with 11 billion parameters, designed for real-time bidirectional voice conversation. It supports simultaneous audio stream transmission and reception while maintaining dynamic state across sessions.
Timeline (last 90 days)
NVIDIA Nemotron VoiceChat 11B is a full-duplex voice model with a parameter scale of 11B.
In the NVIDIA reference technology stack, every 160 milliseconds of audio is decomposed into thousands of tiny GPU operations, causing the GPU to be idle and waiting approximately 75% of the time.
All Facts (2)
NVIDIA Nemotron VoiceChat 11B is a full-duplex voice model with a parameter scale of 11B.
50%UnverifiedIn the NVIDIA reference technology stack, every 160 milliseconds of audio is decomposed into thousands of tiny GPU operations, causing the GPU to be idle and waiting approximately 75% of the time.
50%