This paper analyzes the gender representation in four major corpora of French\nbroadcast. These corpora being widely used within the speech processing\ncommunity, they are a primary material for training automatic speech\nrecognition (ASR) systems. As gender bias has been highlighted in numerous\nnatural language processing (NLP) applications, we study the impact of the\ngender imbalance in TV and radio broadcast on the performance of an ASR system.\nThis analysis shows that women are under-represented in our data in terms of\nspeakers and speech turns. We introduce the notion of speaker role to refine\nour analysis and find that women are even fewer within the Anchor category\ncorresponding to prominent speakers. The disparity of available data for both\ngender causes performance to decrease on women. However this global trend can\nbe counterbalanced for speaker who are used to speak in the media when\nsufficient amount of data is available.\n