Nvidia released Nemotron 3.5 Transcribe, an open AI model that transcribes, translates, and answers questions about audio in a single pass. It processes up to 100 hours of speech across 40 languages on hardware ranging from edge chips to data centers.
FireBrief
FireBrief