NVIDIA Releases Nemotron 3.5 ASR: Live Demo with Multilingual Streaming and Diarization

Short summary

NVIDIA released Nemotron 3.5 ASR, a new automatic speech recognition model with multilingual streaming, word boosting, and speaker diarization capabilities. Sam Witteveen provides a detailed walkthrough with live demonstrations, model quantization options, and hands-on code examples. Available on Hugging Face with full documentation and GitHub tutorials.

• •Nemotron 3.5 ASR supports multilingual streaming, word boosting, and speaker diarization
• •Includes quantized model variants (0.6B) for edge deployment
• •Complete demo walkthrough with code examples and model card documentation

Generated with AI, which can make mistakes.