Picture published by the creator on Hugging Face.
NVIDIA-Nemotron-3.5-Transcribe-30B-A4B: Transcribe hour-long recordings (34B)
Added · Updated
Made by NVIDIA
View profile on Hugging Face (opens in a new tab)Transcribe recordings up to an hour long in 40 languages, with speaker labels and timestamps, or translate speech.
For developers building speech-to-text tools.
Model details
Original model nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B
Input Audio Output Text
Ways to use it Browser demo · Run locally
Role in this family: Fine-tuned version.
nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 (opens in a new tab)What it does
You give it audio plus a text instruction, and optional context, and it writes a text response. A hybrid Mixture of Experts design (only some parts run for each step) uses about 4.3B of its 34B parameters, including the audio encoder. It handles transcription, translation, speaker-labelled transcripts, timestamps, language identification and questions about the audio.
How to use it
Load the weights from the Hugging Face page with NVIDIA's NeMo toolkit; the model card's quick start and prompt guide show the steps. Lower-precision FP8 and NVFP4 versions are also published. For transcription and translation, use temperature (randomness setting) 0 with reasoning off; for questions about audio, temperature 0.6 with reasoning on.
Common use cases
- Upload a one-hour meeting recording and get a transcript showing which speaker said each line.
- Give it a recorded interview with an instruction to translate the speech, and get the translated text back.
- Upload a lecture recording, ask a question about what was said, and get a text answer.
Can I use this?
Requirements The weights download from Hugging Face without an access request. Install NVIDIA's NeMo toolkit, then follow the model card's quick start; the inference setup section lists the prompts for each task.
Caveat Transcription covers 40 languages; check the model card's language support matrix to confirm your language before relying on it.
Worth knowing It works on recorded audio (offline), up to 1 hour long. Licensed under the OpenMDW License 1.1; the model card says it is ready for commercial or non-commercial use.
Why it's here
- 3,105 people have liked it on Hugging Face.
- It was downloaded 67,230 times in the last 30 days.
Numbers as of 8 November 2026, from BuildTube’s latest daily refresh.
Behind it
View model pageBuildTube has not run or verified this project. Everything above is written from what the creator published.
Made this?
Quick questions
- What is NVIDIA-Nemotron-3.5-Transcribe-30B-A4B?
- You give it audio plus a text instruction, and optional context, and it writes a text response.
- What can I use NVIDIA-Nemotron-3.5-Transcribe-30B-A4B for?
- Upload a one-hour meeting recording and get a transcript showing which speaker said each line; Give it a recorded interview with an instruction to translate the speech, and get the translated text back; Upload a lecture recording, ask a question about what was said, and get a text answer
- Can I run NVIDIA-Nemotron-3.5-Transcribe-30B-A4B on my computer?
- It runs on your own hardware with a model runtime. The model card mentions Transformers, SGLang, vLLM; check its setup and hardware notes.
- What do I need to use NVIDIA-Nemotron-3.5-Transcribe-30B-A4B?
- One NVIDIA H100 GPU with 80 GB, per the model card's deployment line, and the NeMo software library.
- What license does NVIDIA-Nemotron-3.5-Transcribe-30B-A4B use?
- Licensed under the OpenMDW License 1.1; the model card says it is ready for commercial or non-commercial use.
More from this week
- Handy Dictate into any text field with offline speech-to-text
Handy lets you press a shortcut, speak, and see your words appear in any text field, transcribed on your own computer.
- NVKE Run managed Kubernetes clusters on your own NVIDIA GPU machines
Set up a service that creates and manages separate Kubernetes clusters on your own NVIDIA GPU servers.
- P-H3-Image-Comfy Make and edit 8-megapixel images in ComfyUI
Turn short text requests into 8-megapixel images and edit pictures on your own graphics card.
Explore by topic
Post this
Download share imageFor makers
Made this? Show it’s on BuildTube — add the badge to your README or embed the card on your site.
README badge (Markdown)
[](https://buildtube.app/p/hf-nvidia-nvidia-nemotron-3-5-transcribe-30b-a4b/)Embed card (HTML)
<iframe src="https://buildtube.app/embed/hf-nvidia-nvidia-nemotron-3-5-transcribe-30b-a4b/" width="560" height="240" style="border:0;max-width:100%" loading="lazy" title="NVIDIA-Nemotron-3.5-Transcribe-30B-A4B on BuildTube"></iframe>