Whisper

Whisper is OpenAI's multilingual speech-recognition (STT) model open-sourced in September 2022. It was trained on 680,000 hours of supervised web audio and supports transcription in 99 languages, language detection, and English translation. The open release seeded a broad ecosystem from podcast tooling to meeting assistants, and projects like whisper.cpp brought it to on-device inference. The large-v3-turbo release in 2024 marked another step forward on the speed/accuracy curve.