Whisper
OpenAI's open-source speech recognition model — the benchmark for multilingual STT
OpenAI's open-source speech recognition model; the benchmark for multilingual speech-to-text.
Text-to-speech, speech-to-text, and voice cloning, open source.
Key features
- General-purpose speech recognition trained on diverse audio
- Multilingual recognition, speech translation, language ID
- Transformer seq2seq replacing traditional speech pipelines
- Six model sizes from tiny to turbo with speed/accuracy tradeoffs
Good for
- Transcribing and translating speech from audio files
- Developers needing robust multilingual ASR
Stack & tags
PythonPyTorchspeech-recognitionasrmultilingualtranslationopenaiopen-weightstranscription
Quick start
pip install -U openai-whisperFrom the project README on GitHub.
Screenshots
Screenshots from the project’s GitHub README.
Similar open-source projects
GPT-SoVITS
★ 62,271TTS/cloning trainable on 1 minute of audio — a legendary Chinese-community project
Real-Time-Voice-Cloning
★ 60,133The classic 5-second voice cloning project — the "founding father" of voice cloning