From Ici Première to Ici Musique Classique, from national broadcasts to regional broadcasts, from podcasts to major series, Radio-Canada produces several dozen hours of content per day. In order to improve our content offering and optimize our operations, transcription is becoming a central element in our work. This allows us on the one hand to make our content accessible to deaf and hard of hearing people and on the other hand to apply different text-based machine learning methods, such as automatic summarization. Our presentation explains the process of implementing a transcription tool at Radio-Canada, from proof of concept to production.