- Published on
Whisper is a general-purpose speech recognition model that can perform multilingual speech recognition, speech translation, and language identification. It is trained on a large dataset of diverse audio and uses a Transformer sequence-to-sequence model.