Export
The link always opens the latest version of this page.
Ask an AI about it
Open in ChatGPTA new chat that reads this entry from the site Open in ClaudeThe same, in ClaudeShare the link instead of a file, and they will always see the latest version.
Files are a snapshot of today. The live page has anything newer.
NVIDIA Nemotron 3.5 Transcribe 30B A4B
Open-weight offline speech-to-text model with 34B parameters supporting transcription, translation, and audio QA across 40 languages
- Category
- speech recognition model 1 source
NVIDIA Nemotron 3.5 Transcribe 30B A4B is an open-weight speech recognition model with 34 billion parameters designed for offline speech-to-text processing 1.
The model supports transcription, translation, and audio question-answering across 40 languages 1. It handles context lengths up to 65,536 tokens and can process audio inputs up to 1 hour in duration 1.
Performance metrics include a word error rate (WER) of 4.71% on English benchmarks, 10.98% on long-form audio, and 4.62% on multilingual content 1.
The model is optimized for deployment on NVIDIA H100 hardware 1.
Facts 2
| Attribute | Value | Confidence | Sources | Cited |
|---|---|---|---|---|
| Vendor | Nvidia | low | 1 | 1 |
| Category | speech recognition model | low | 1 | 1 |
Version 2
- mean WER 4.71 English, 10.98 long-form, 4.62 multilingual
- supports 40 languages, context length 65,536 tokens, up to 1 hour audio