NVIDIA发布开源语音模型Nemotron 3.5，支持多语种转录与问答
NVIDIA is releasing an open speech model that can transcribe a meeting, label wh…
NVIDIA开源的30B参数语音大模型，性能对标商业级，适合需要本地部署或定制多语种会议转录的团队直接落地。
NVIDIA is releasing an open speech model that can transcribe a meeting, label who said what, translate it and answer questions about the recording:
NVIDIA 正在发布一款开源语音模型，它可以转录会议内容、标注发言者、进行翻译并回答关于录音的问题：
Nemotron 3.5 Transcribe supports: - Transcription in 40 languages - Translation between English and 31 languages - Up to eight speakers, with speaker labels in 13 languages - Recordings up to an hour long
Nemotron 3.5 Transcribe 支持： - 40 种语言的转录 - 英语与 31 种语言之间的翻译 - 最多八位说话人，支持 13 种语言的说话人标签 - 最长一小时的录音
You can give it names and technical terms to guide the transcript, or ask questions about the audio. Audio Q&A is trained in English.
你可以提供名称和技术术语以指导转录结果，或者询问有关音频的问题。音频问答功能使用英语训练。
It has roughly 30B parameters, with about 4B active per token, and can run on your own NVIDIA GPU.
该模型拥有约 300 亿参数，每个 token 约有 40 亿活跃参数，可以在你自己的 NVIDIA GPU 上运行。
According to NVIDIA's launch materials, it ranks first among open-weight models on Hugging Face's English Open ASR leaderboard and leads VoiceArena's 14-system test of transcription with speaker attribution.
根据 NVIDIA 的发布资料，它在 Hugging Face 的英文开源自动语音识别（ASR）排行榜上位居开源权重模型之首，并在 VoiceArena 针对 14 个系统的带说话人归属的转录测试中领先。
Built for recorded audio. Live streaming uses a separate model.
专为录制音频设计。直播流媒体使用单独的模型。
Super cool to see!
看到真是太酷了！
Model: https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B Demo: https://huggingface.co/spaces/nvidia/nemotron-transcribe Blog: https://huggingface.co/blog/nvidia/nemotron-transcribe-granary/
模型：https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B 演示：https://huggingface.co/spaces/nvidia/nemotron-transcribe 博客：https://huggingface.co/blog/nvidia/nemotron-transcribe-granary/
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力