Whisper
Multilingual speech recognition and translation model with a CLI
What it is
Whisper is a family of open speech recognition models, from 39M to 1.5B parameters, that transcribe audio in many languages, translate speech into English, and identify the spoken language. It ships as a Python package and a command-line tool that takes common audio files and returns text with timestamps. Developers and researchers use it to add transcription to their own software or to batch-convert recordings into text locally.