Nvidia launches AI model converting speech to text supporting 40 languages
Nvidia announced the launch of the Nemotron 3.5 Transcribe model for speech-to-text conversion, supporting 40 languages and processing recordings up to an hour long
Translated by Qiyas from الوسط — Original source
US-based company NVIDIA announced the launch of a new artificial intelligence model named “Nemotron 3.5 Transcribe,” designed to convert audio recordings into written text. The model supports 40 languages and allows developers to use it for free with open weights. The company clarified that the new model supports translation between English and 31 other languages, can process audio recordings up to one hour in length, and can distinguish among up to eight speakers in a single recording. It added that the model targets the development of applications for transcribing audio recordings and medical conversations into written text, with the ability to recognize specialized technical terminology across various fields. The model comprises 30 billion parameters and can run on a single chip across a variety of computing platforms, enabling its use in diverse applications without the need for multiple computing systems.
