AI News AI资讯 13d ago Updated 13d ago 更新于 13天前 51

NVIDIA Releases NemotronLabs VoiceChat 11B: An Open Full-Duplex Speech-to-Speech Model with ~450 ms Turn-Taking and Live Tool Calling 英伟达发布 NemotronLabs VoiceChat 11B:一款开源全双工语音到语音模型,具备约 450 毫秒对话切换和实时工具调用能力

NVIDIA released NemotronLabs VoiceChat 11B, an open 11B-parameter end-to-end speech-to-speech model that eliminates the traditional ASR→LLM→TTS cascade in favor of a unified full-duplex network The model achieves 448 ms smooth turn-taking latency and a 1.00 take-over rate at 480 ms, enabling natural barge-in conversation where users can interrupt mid-turn It is the first open full-duplex model to support tool calling via a separate side-channel output with operator-defined "on-hold" messages to NVIDIA发布NemotronLabs VoiceChat 11B,一个开放的11B参数端到端语音到语音模型,支持实时全双工对话 模型采用统一网络架构替代传统ASR→LLM→TTS级联方案,实现448ms的低延迟对话切换 首个支持工具调用的开源全双工模型,通过独立输出通道处理<TOOLCALL>脚本并支持"on-hold"消息 混合Mamba/Transformer架构,训练数据约550k小时音频,在Full-Duplex-Bench 1.0上表现优异 当前仅适用于研究试点,存在2分钟音频上下文限制、多轮对话退化、自说自话等问题,需80GB VRAM GPU

75
Hot 热度
72
Quality 质量
70
Impact 影响力

Analysis 深度分析

TL;DR

  • NVIDIA released NemotronLabs VoiceChat 11B, an open 11B-parameter end-to-end speech-to-speech model that eliminates the traditional ASR→LLM→TTS cascade in favor of a unified full-duplex network
  • The model achieves 448 ms smooth turn-taking latency and a 1.00 take-over rate at 480 ms, enabling natural barge-in conversation where users can interrupt mid-turn
  • It is the first open full-duplex model to support tool calling via a separate side-channel output with operator-defined "on-hold" messages to prevent conversational silence during API calls
  • Despite permissive OpenMDW-1.1 licensing, NVIDIA labels the checkpoint "research purposes only" due to documented failure modes including a 2-minute audio context ceiling, multi-turn degradation into gibberish, and dropped transcription words
  • The model requires a single GPU with at least 80 GB VRAM (A100/H100/RTX 6000 Pro/B200) and currently has no hosted API or inference provider

Why It Matters

This release represents a significant architectural shift in conversational AI, demonstrating that unified speech-to-speech models can compete with cascaded ASR-LLM-TTS pipelines on latency while enabling full-duplex interaction—a capability previously limited to closed proprietary systems. For AI practitioners, it provides the first open benchmark for evaluating full-duplex tool-calling voice agents, though the research-only designation and hardware requirements signal that production readiness remains a work in progress.

Technical Details

  • Architecture: Hybrid Mamba/Transformer combining a Fast Conformer speech encoder (from Nemotron-Speech-Streaming-En-0.6b) for continuous 16 kHz audio encoding, the Nemotron Nano v2 LLM backbone for audio-to-text token prediction, an NVIDIA TTS decoder and codec for 22.05 kHz speech generation, and a separate output channel dedicated to <TOOLCALL> scripts
  • Training: Approximately 550k hours of audio across real and synthetic corpora, built on SALM-Duplex and Audio Flamingo 3 datasets
  • Tool-calling mechanism: Tool calls are emitted on a side channel as <TOOLCALL> blocks; the system returns results via <TOOL_RESPONSE> blocks, with operator-defined on-hold messages spoken during API execution to maintain conversational flow
  • Performance benchmarks: Ranks #2 among open full-duplex models on VoiceBench and Full-Duplex-Bench 1.0; AU Harness BFCL-v3 spoken tool calling scores 56.1% average (58.5% simple, 62.5% multiple, 42.5% parallel); Full-Duplex-Bench v3 shows 82.5% tool selection, 44.2% argument accuracy, 33% pass@1
  • Constraints: Maximum 5 tools per session, no reliable simultaneous multi-tool calls, no user interruption during tool execution, ASCII-only system prompts and tool responses required

Industry Insight

  • The 448 ms turn-taking latency and full-duplex barge-in capability set a new open-source baseline for real-time voice agents, accelerating competitive pressure on proprietary systems like Google's Gemini Voice and Anthropic's voice features that previously dominated this space
  • The explicit failure modes (context ceiling, multi-turn degradation, transcription drops) highlight that end-to-end speech models still struggle with long-context stability and error propagation—teams piloting this model should implement external monitoring and fallback to cascaded pipelines for production-critical applications
  • The 80 GB GPU requirement and absence of a hosted API create a barrier to entry that favors well-funded organizations and research labs; expect inference optimization efforts (quantization, distillation) to emerge quickly, and anticipate third-party providers building managed serving layers within months

TL;DR

  • NVIDIA发布NemotronLabs VoiceChat 11B,一个开放的11B参数端到端语音到语音模型,支持实时全双工对话
  • 模型采用统一网络架构替代传统ASR→LLM→TTS级联方案,实现448ms的低延迟对话切换
  • 首个支持工具调用的开源全双工模型,通过独立输出通道处理脚本并支持"on-hold"消息
  • 混合Mamba/Transformer架构,训练数据约550k小时音频,在Full-Duplex-Bench 1.0上表现优异
  • 当前仅适用于研究试点,存在2分钟音频上下文限制、多轮对话退化、自说自话等问题,需80GB VRAM GPU

为什么值得看

NVIDIA此举标志着端到端语音交互模型的重要突破,将全双工对话与工具调用能力结合,为实时语音Agent提供了新的技术路径。尽管目前存在稳定性问题,但其开源策略和架构设计为行业研究提供了宝贵参考。

技术解析

  • 架构设计:采用混合Mamba/Transformer架构,由Fast Conformer语音编码器、Nemotron Nano v2 LLM骨干、NVIDIA TTS解码器和codec组成,新增独立工具调用输出通道
  • 性能指标:在Full-Duplex-Bench 1.0上实现448ms平滑对话切换延迟,用户打断接管率1.00(480ms),语音工具调用BFCL-v3平均56.1%
  • 训练数据:使用约550k小时音频(真实+合成语料),基于SALM-Duplex和Audio Flamingo 3构建
  • 工具调用机制:通过块发送工具调用,块返回结果,支持操作员定义的"on-hold"消息避免API调用时的沉默间隙
  • 部署限制:需单GPU 80GB VRAM(A100/H100/RTX 6000 Pro/B200),无托管API,最多支持5个工具/会话,系统提示和工具响应需为ASCII-only

行业启示

  • 端到端模型趋势:NVIDIA验证了统一网络替代级联架构的可行性,未来语音交互领域可能加速向端到端方案演进
  • 全双工+工具调用结合:首次将实时打断能力与工具调用结合,为客服、车载助手等场景提供了更自然的交互范式
  • 研究到生产的鸿沟:模型明确标注"研究用途",暴露了语音Agent在长上下文、稳定性方面的技术挑战,提醒行业需关注基础模型可靠性问题

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

Open Source 开源 Speech 语音 Conversational AI 对话系统 LLM 大模型 Agent Agent