Yesterday OpenAI published a new paper about multilingual speech recognition and, to everyone’s surprise, they also publicly released the model. Thanks to a mix of curiosity and procrastination, I tried it out in Polish on various domains such as slang, politics and medical text books.