Whisper

OpenAI Whisper is a general-purpose speech recognition model trained on 680,000 hours of multilingual and multitask supervised data. It supports transcription in over 50 languages, translation to English, language identification, and voice activity detection, making it one of the most versatile open speech recognition systems available.

The model uses a simple encoder-decoder Transformer architecture trained on diverse audio quality, accents, and background noise conditions, resulting in robust performance across real-world scenarios. It approaches human-level accuracy on many transcription benchmarks, particularly for English.

Whisper has become the foundation for countless speech recognition applications, from automated transcription services to voice assistants, and is available in multiple model sizes for different speed-accuracy tradeoffs.