Whisper

E20825

Whisper is an open-source automatic speech recognition system by OpenAI that transcribes and translates spoken language with high accuracy across many languages.

Jump to: Surface forms Statements Referenced by

Observed surface forms (1)

Surface form Occurrences
Whisper model 1

Statements (69)

Predicate Object
instanceOf automatic speech recognition system
open-source software
developer OpenAI
feature robustness to accents
robustness to background noise
robustness to technical language
hasModelVariant base
large
medium
small
tiny
input audio
modelType encoder-decoder transformer
openSourceRepository https://github.com/openai/whisper
organization OpenAI
output text
programmingLanguage C++
Python
provides pretrained models of multiple sizes
releaseDate September 2022
softwareLicense MIT License
supports multilingual speech recognition
multilingual speech translation
segment-level language detection
timestamped transcription
transcription of spoken language
translation of spoken language
supportsLanguage Arabic
Bulgarian
Cantonese
Chinese
Czech
Danish
Dutch
English
Finnish
French
German
Greek
Hebrew
Hindi
Hungarian
Indonesian
Italian
Japanese
Korean
Malay
Norwegian language
surface form: Norwegian

Polish
Portuguese
Romanian language
surface form: Romanian

Russian
Spanish
Swahili language
surface form: Swahili

Swedish
Tagalog
Thai
Turkish language
surface form: Turkish

Ukrainian
Vietnamese
task language identification
speech recognition
speech translation
speech-to-text transcription
trainingData multilingual and multitask supervised data collected from the web
useCase building voice interfaces
captioning videos
transcribing meetings
transcribing podcasts

Referenced by (3)

Full triples — surface form annotated when it differs from this entity's canonical label.

OpenAI developed Whisper
this entity surface form: Whisper model