Instructions to use nvidia/nemotron-3.5-asr-streaming-0.6b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- NeMo
How to use nvidia/nemotron-3.5-asr-streaming-0.6b with NeMo:
import nemo.collections.asr as nemo_asr asr_model = nemo_asr.models.ASRModel.from_pretrained("nvidia/nemotron-3.5-asr-streaming-0.6b") transcriptions = asr_model.transcribe(["file.wav"]) - Transformers
How to use nvidia/nemotron-3.5-asr-streaming-0.6b with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="nvidia/nemotron-3.5-asr-streaming-0.6b")# Load model directly from transformers import AutoProcessor, AutoModel processor = AutoProcessor.from_pretrained("nvidia/nemotron-3.5-asr-streaming-0.6b") model = AutoModel.from_pretrained("nvidia/nemotron-3.5-asr-streaming-0.6b", device_map="auto") - Inference
- Notebooks
- Google Colab
- Kaggle
Malayalam Support?
why no Indic languages yet? please support Malayalam, eagerly waiting...
Malayalam is not in the current supported set, but the model's tokenizer and architecture support adding new languages through fine-tuning.
The fine-tuning recipe is here: https://huggingface.co/blog/nvidia/fine-tuning-nemotron-35-asr
You'd need a Malayalam speech dataset with punctuated, properly-cased transcripts. Common Voice and FLEURS both have Malayalam splits. For a community example of adding a new language, see this Swahili adaptation: https://huggingface.co/Tonykip/nemotron-3.5-swahili-streaming-asr
We're tracking demand for Indic languages. Hindi is already transcription-ready, and expanding to more Indic locales is on the roadmap.