What is Whisper?
Whisper is a robust AI-powered speech recognition tool that uses large-scale weak supervision. It is a general-purpose model that can perform multilingual speech recognition, speech translation, and spoken language identification. It is based on a sequence-to-sequence model that allows for joint representation of sequence tokens and prediction decoding. It offers five available model sizes with varying speed and accuracy tradeoffs. It is open-source under the MIT license.
⭐ Whisper Core features
- ✔️ Speech recognition
- ✔️ Speech translation
- ✔️ Spoken language identification
- ✔️ Sequence-to-sequence model
- ✔️ Joint representation of sequence tokens and prediction decoding