
Whisper
Open-source speech recognition
Whisper is OpenAI's open-source speech recognition model that approaches human-level accuracy. Transcribe and translate audio in 99 languages. Robust to accents, background noise, and technical language. Run locally for privacy or use OpenAI's API for convenience. Multiple model sizes balance accuracy and speed. The speech recognition model that finally makes transcription reliable enough to trust.











