Acoustic front-end processing for communication systems

Summary form only given. As communication systems have become more mobile and portable, we now have situations where audio communication in difficult acoustic environments is common. Speech coders at low bit rates tend to have problems with non-speech signals that are typically found in noisy acoustic environments. As a result, there can be degradation in the perceived audio quality for low-bit rate coders in noisy environments. One solution that has not been well studied, is the interaction of the acoustic front-end with speech coders in these difficult environments. Clearly an understanding of the end-to-end communication channel including the acoustic front-end will be necessary in order to optimize signal coding. Another area where acoustic front-end processing will play a large role in speech and audio coding is Internet protocol (IP) based communication networks. IP networks enable us to "easily" deliver wider bandwidth and multiple channels of audio for more realistic and transparent telecommunication. Sound transduction for hands-free teleconferencing has audio quality issues similar to the mobile communication problem. This article concentrates on some of the work that we have been doing at Bell Labs on the hands-free telecommunication problem with an emphasis on microphone arrays and acoustic echo cancellation. Finally, some new areas for possible collaboration between the speech coding and electroacoustics communities to improve mobile and hands-free communication are suggested.