Nvidia launches open Nemotron 3.5 AI model for audio transcription and translation

Nvidia Corp. has introduced Nemotron 3.5 Transcribe, an open artificial intelligence model designed for multilingual audio transcription and translation. The model consolidates multiple tasks—speech detection, transcription, speaker identification, and question answering—into a single system, streamlining workflows for developers and enterprises. Nvidia also released an open data pipeline to fine-tune the model for specific use cases, emphasizing adaptability. This move targets industries like media, customer service, and accessibility, where accurate, scalable audio processing is critical. By open-sourcing the model and pipeline, Nvidia aims to accelerate adoption and innovation in speech AI, competing with proprietary offerings from tech giants. The release underscores the growing importance of multimodal AI in handling real-world audio data, with potential applications in global communication and content creation. Nvidia's push into open models could reshape the landscape for developers seeking cost-effective, customizable solutions.