The enhancement of speech intelligibility in noise is still the main subject in hearing aid research. Based on the advanced results obtained with the hearing glasses, in the present research the speech intelligibility is even further improved by the application of binaural post-filters. The functionalities of these filters are related to the principles of the auditory scene analysis. A statistical analysis of binaural cues in noise at the output of different hearing aids, the utilization of a Bayesian classifier in the source separation process and an evolutionary optimization against binaural models of speech intelligibility provides a comprehensive understanding for the utilization of binaural post-filters in adverse environments. As the listening ease and a fair amount of speech quality are mandatory in speech enhancement, tradeoffs between speech intelligibility and quality were studied in terms of the preservation of natural binaural cues and the suppression of musical noise.