NVIDIA has introduced Nemotron 3.5 Transcribe, an open automatic speech recognition model covering 40 languages. Announced in a Hugging Face blog post, it aims to make multilingual ASR more accessible to developers.