Skip to content

Latest commit

 

History

History
 
 

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 

Whisper model

Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation.

Whisper was proposed in the paper Robust Speech Recognition via Large-Scale Weak Supervision by Alec Radford et al. from OpenAI. The original code repository can be found here.

alt text

Model lists

There are ten variations of the models. They are available on the Hugging Face Hub. The checkpoints are summarised in the following table with links to the models on the Hub:

Size Parameters English-only Multilingual
tiny 39 M
base 74 M
small 244 M
medium 769 M
large 1550 M x
large-v2 1550 M x