A Real-Time Dual-Channel Speech Reinforcement System for Intra-Cabin Communication

In order to facilitate the communication experience in a cabin (as among passengers located at distant seats in big vehicles), suitable systems involving the presence of microphones, amplifiers and loudspeakers are needed. However when the sound is acquired and reproduced within the same acoustic environment, instability problems can arise due to acoustic feedback. The main objective of Speech Reinforcement techniques consists in reducing the occurrence of feedback effects, thus allowing for a comfortable intra-cabin communication. In this work we focus on the PEM-AFROW algorithm, an effective technique for acoustic feedback control, recently appeared in the literature. Moreover a suppressor filter is included within the feedback loop in order to improve the overall system stability and increase its inherent robustness to higher gain values. The overall algorithmic scheme is able to deal with the DualChannel communication case study, i.e. in presence of two communicating speakers and with the speech information flowing in both directions. This means that, in contrast to the SingleChannel case study, echo paths, introduced by the double microphone and loudspeaker, must be considered. In order to keep the latencies low and allow a real-time processing, the partitioned block frequency domain adaptive filter (PB-FDAF) algorithm has been adopted. Voice Activity and Double Talk Detectors have been also included into the algorithmic framework. Performed computer simulations in various acoustic conditions have shown the effectiveness of the approach.