Submit ProductSign in
Intelligent Transcription with a Speech LLM
Grounded in available product and source data
Nemotron 3.5 Transcribe is a speech-augmented language model for intelligent transcription, where the right convention is chosen per request rather than fixed at training time. One model handles verbatim, written-form and contextual output.
It couples a speech encoder and a frozen diarization model with a hybrid mixture-of-experts LLM, adds speaker attribution and word-level timestamps, and supports transcription, translation and spoken language identification across many languages.
