NVIDIA Releases Nemotron 3.5 Transcribe: An Open 30B MoE Model for 40-Language Transcription with ~4B Active Parameters

TL;DR

NVIDIA released Nemotron 3.5 Transcribe, an open-weights speech LLM with 40-language transcription capabilities and improved accuracy metrics. This model marks a step forward in multilingual speech recognition technology.

Detailed Summary

NVIDIA released Nemotron 3.5 Transcribe, an open-weights speech LLM with 4.3B active parameters capable of transcribing 40 languages and achieving a 4.71% mean WER on the English Open ASR Leaderboard. The model includes speaker labels and word timestamps. This release broadens NVIDIA's offerings in natural language processing, potentially enhancing multilingual speech recognition technologies.

Key Points

• • Nemotron 3.5 Transcribe is an open-weights speech LLM.
• • It has 4.3B active parameters and transcribes 40 languages.
• • The model adds speaker labels and word timestamps.
• • It achieves a 4.71% mean WER on the English Open ASR Leaderboard.