NVIDIA releases a 2B-parameter audio-native model with open weights that handles transcription, translation, sound recognition, audio Q&A, text-to-speech, and full speech-to-speech in one model. The post says the model "hears the world" rather than only words, and links a Hugging Face demo.
NVIDIA dropped an 2B audio-native model that doesn’t just hear words it hears the world.
Transcription, translation, sound recognition, audio Q&A, TTS, and full speech-to-speech, all in one model.
Open weights Play with it here: https://huggingface.co/spaces/nvidia/Nemotron-Labs-Audex
Source: Md Ismail Šojal 🕷️ · x.comPublished · added here
