
NVIDIA Beats Whisper with Parakeetv2 - Exploring NVIDIA's Latest ASR System
Sam Witteveen via YouTube
Overview

Coursera Plus Annual Sale:
All Certificates & Courses 50% Off!
Grab it
Explore NVIDIA's latest open-weight Automatic Speech Recognition (ASR) system, Parakeetv2, which outperforms OpenAI's Whisper model. This 10-minute video tutorial demonstrates the capabilities of NVIDIA's new ASR technology, walking through the Hugging Face implementation, available spaces, research papers, and GitHub repository. Follow along with a practical Colab demonstration that showcases Parakeetv2 in action. The video includes references to Distil-Whisper and MLX versions of Whisper for comparison. Access the provided Colab notebook, Hugging Face model, and Spaces links to experiment with this advanced speech recognition system yourself.
Syllabus
00:00 Intro/ OpenAI Whisper Blog
00:42 Distil-Whisper Hugging Face
00:53 Whisper MLX Version
00:56 Hugging Face NVIDIA Parakeet
02:13 Hugging Face Spaces Parakeet
02:49 Papers
03:27 NVIDIA Parakeet Github
03:58 Parakeet V2 Colab Demo
Taught by
Sam Witteveen