Nvidia 发布开放权重语音转写模型 Nemotron 3.5 Transcribe，支持 40 种语言
Voice AI in vogue: Nvidia releases new open transcription model
- 1Nvidia 发布免费开放权重语音转写模型 Nemotron 3.5 Transcribe，支持最多 40 种语言
- 2官方称支持 31 种语言与英语互译、最长 1 小时录音、区分 8 位说话人，并称基准超对手
- 330B 参数版本宣称可在 Jetson AGX Thor 到 Blackwell 单芯片运行，但仅支持录音、不支持实时流
- 4同步发布开源语音数据集 Granary 2，面向开发会议记录与医疗转写应用的开发者
Nvidia 以免费开放权重放出语音转写模型，把会议与医疗转写这类成熟场景的技术和成本门槛进一步拉低。
- 影响：会议记录、医疗转写等已有付费场景的应用成本结构可能改变，开发商可评估用开放权重模型替代按量计费的转写 API；仅支持录音不支持实时流，实时客服类场景暂不在覆盖范围；免费开放权重也降低了同类应用的竞争门槛。
- 参考：可考虑用业务自己的录音数据实测对比现有转写供应商；试点宜选会议纪要或病历转写这类录音场景；行业术语定制依赖 Granary 数据集与 NeMo 工具链，值得先评估工程投入。
- 依据：40 语言转写、31 语言互译、单芯片运行与基准领先均为 Nvidia 官方口径，暂无独立验证；仅支持录音是官方说明的功能边界。
In Brief
Posted:
8:00 AM PDT · October 28, 2026
Nvidia on Wednesday released Nemotron 3.5 Transcribe, a free, open-weight model that transcribes audio in up to 40 languages. It also released version 2 of Granary, its open-source speech dataset.
Nvidia boasts that Nemotron 3.5 can translate 31 languages to and from English, handle recordings up to an hour long, and tell up to eight speakers apart. It claims to beat rivals on benchmarks, including Hugging Face’s Open ASR Leaderboard.
These tools target developers building apps for business meetings and medical transcription, two of AI’s most established footholds in the workplace. Nvidia says developers can use the model, the updated Granary dataset, and tools like its NeMo Speech model to customize apps to recognize an industry’s jargon in each language. One drawback: Transcribe works with recorded audio, not live streams.
Still, the company promises Transcribe, at 30 billion parameters, can run on a single chip, from the Jetson AGX Thor robotics controller and DGX Spark chip powering new laptops, to Blackwell datacenter chips. It joins Nvidia’s open-weight voice models, including the super small yet powerful Parakeet (1.1 billion parameters).
Nvidia’s latest voice entry comes during busy period of AI voice releases: Mistral shipped Voxtral 3 earlier this month; ElevenLabs launched Realtime for instant speech in voice agents; and Salesforce released Launa-Omni, its voice model.
来源：TechCrunch · AI · techcrunch.com