This paper focuses on recognizing Arabic text embedded in videos. The proposed methods proceed without applying any prior pre-processing operations or character segmentation. Difficulties related to the video or text properties are faced using a learned robust representation of the input text image. This is performed using deep auto-encoders and Convolutional Neural Networks. Features are computed using a multiscale sliding window scheme. A connectionist recurrent approach is then used. It is trained to predict correct transcriptions of an input image from the associated sequence of features. Proposed methods are extensively evaluated on a large database of Arabic TV channels videos and compared to existing solutions.