CochleaNet: A Robust Language-independent Audio-Visual Model for Speech Enhancement

•Audiovisual (AV) system integrates noisy audio and speakers lips to enhance speech.•AV system trained on synthetical mixtures works in real noisy social environments.•AV system works on unseen languages, speakers and noises without SNR estimation.•AV system in compared with state-of-the-art speech enhancement models.•Analysis reveal the role of visual cues in the superior performance of AV System.