NemotronLabs VoiceChat: An Open Full-duplex Speech-to-Speech Model with Tool Calling Capabilities
NemotronLabs VoiceChat is an open full-duplex speech-to-speech model that calls tools during real-time conversation.
NemotronLabs VoiceChat combines a streaming speech encoder and decoder-only language model with parallel text and function-call outputs, an RNN-T transcription branch, and a streaming TTS decoder. On Full-Duplex-Bench 1.0 it reports the lowest pause-handling takeover among open-weight systems, 100% takeover after interruptions, and 4.33/5 response quality. It resumes after backchannels in 93% of Full-Duplex-Bench 1.5 cases, scores 55.1 on VoiceBench, and reaches 82.5% tool-selection F1 on Full-Duplex-Bench 3.0, while argument accuracy and end-to-end execution still lag.
- Open streaming speech-to-speech model with native tool calling.
- Lowest pause-handling takeover on Full-Duplex-Bench 1.0 among open-weight systems.
- 100% interruption takeover and 4.33/5 post-interruption response quality.
- VoiceBench score 55.1; 82.5% tool-selection F1 on Full-Duplex-Bench 3.0.
- Argument accuracy and end-to-end tool execution remain weaker.
Full article177 words · extracted from arxiv.org · click to collapse
We introduce NemotronLabs VoiceChat, an open full-duplex speech-to-speech model with native tool-calling capabilities. NemotronLabs VoiceChat combines a streaming speech encoder and decoder-only language model with parallel specialized output streams for agent text and structured function calls, an auxiliary RNN-T branch for incremental user transcription, and a streaming TTS decoder. This design enables the model to listen, transcribe, reason, invoke tools, and speak within a unified streaming architecture while preserving the temporal behavior required for natural conversation. On Full-Duplex-Bench 1.0, NemotronLabs VoiceChat achieves the lowest pause-handling takeover rates among evaluated open-weight systems, 100\% takeover following user interruptions, and a 4.33/5 post-interruption response-quality score. On Full-Duplex-Bench 1.5, it resumes its response after user backchannels in 93\% of cases. NemotronLabs VoiceChat obtains a 55.1 normalized average on VoiceBench and, on Full-Duplex-Bench 3.0 (FDB 3.0), achieves 82.5\% tool-selection F1, while argument accuracy and end-to-end tool execution remain areas for improvement. These results demonstrate that full-duplex interaction, speech recognition and generation, general language capabilities, and external tool use can be integrated in a single open speech-to-speech model without sacrificing real-time conversational behavior.
Text extracted automatically; images, tables and formatting may be missing. Original: https://arxiv.org/abs/2609.21967