Springer, 2005. — 277 p.For convenient human/machine interaction, acoustic front-ends are required which allow seamless and hands-free audio communication. For maximum speech intelligibility and optimum speech recognition performance, interference, noise, reverberation, and acoustic echoes of loudspeakers should be suppressed. Microphone array signal processing is advantageous to single-channel speech enhancement since the spatial dimension can be exploited in addition to the temporal dimension. In this work, joint adaptive beamforming and acoustic echo cancellation with microphone arrays is studied with a focus on the challenges of practical systems. Particularly, the following aspects are efficiently solved, leading to a real-time system, which was successfully used in the real world: High suppression of both strongly time-varying interferers, as, e.g., competing speakers, and slowly time-varying diffuse noise, as, e.g., car noise in passenger cabins of cars, Efficient cancellation of acoustic echoes of multi-channel reproduction systems even in strongly time-varying acoustic conditions with high background noise levels and with limited computational resources, High output signal quality with limited array apertures and limited numbers of microphones because of product design constraints, Robustness against reverberation with respect to the desired signal, moving desired sources, or array imperfections such as position errors or gain and phase mismatch of the microphones. Detailed theoretical analysis and experimental studies illustrate the performance of the system. Audio examples can be found on the web page http://www.wolfgangherbordt.de/micarraybook/. Special focus is put on the reproducibility of the results by giving detailed descriptions of the proposed algorithms and of the parameter settings. The intended audience of this book is both specialists and readers with general knowledge of statistical and adaptive signal processing. For any question or comment, please don’t hesitate to contact the author!Introduction. Space-Time Signals. Optimum Linear Filtering. Optimum Beamforming for Wideband Non-stationary Signals. A Practical Audio Acquisition System Using a Robust GSC (RGSC). Beamforming Combined with Multi-channel Acoustic Echo Cancellation. Efficient Real-Time Realization of an Acoustic Human/Machine Front-End. Summary and Conclusions. A Estimation of Signal-to-Interference-Plus-Noise Ratios (SINRs) Exploiting Non-stationarity. B Experimental Setups and Acoustic Environments. C Notations.
Чтобы скачать этот файл зарегистрируйтесь и/или войдите на сайт используя форму сверху.