open model release

Whisper open-source speech model

OpenAI released Whisper models and inference code for multilingual speech recognition and translation, trained on 680,000 hours of multilingual and multitask supervised web data.