Whisper distil-large-v3

Whisper distil-large-v3 is an automatic speech-recognition model that turns spoken English into text. Sanchit Gandhi, Patrick von Platen, and Alexander Rush introduced Distil-Whisper; this checkpoint distills OpenAI’s Whisper large-v3 into a smaller encoder-decoder model. Distillation trains a student to reproduce useful behavior from a larger teacher. The name can mislead: this is not an OpenAI-released checkpoint merely because it derives from Whisper. The upstream model card describes this release as English speech recognition, so broad multilingual claims are inappropriate. TT-Metal lists batch 1 on either a single n150 Wormhole or a single p150 Blackhole device. The duplicate rows are hardware alternatives for one speech model, not separate models.