Announcement_30

We release Nemotron-3.5-Transcribe, an open-weight speech language model for transcription in 40 languages, with support for contextual biasing, speaker attribution, timestamp prediction, speech translation, and audio question answering.