Single channel speech enhancement by efficient coding

A number of approaches have been proposed for enhancing/extracting a given speech in a noisy environment, in the aim of, for example, enhancing the speech recognition rate. The speech enhancement can be carried out by either using single or multiple channel measurements. Some of those approaches explore the harmonicity of speech, and others make use of the redundancy among the channels, usually by independent component analysis. To enhance one speech from a single channel, we propose here to use the characteristics of speech through the concept of efficient coding, which mimics the way the auditory cortex code information. This is carried out in a supervized fashion. Simulations and real world measurements show that this technique can be used efficiently to enhance a given speech signal.