October 2015
Intermediate to advanced
306 pages
10h 38m
English
Content preview from Robust Automatic Speech Recognition
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
Start your free trial
Chapter 10
Multi-channel processing
Abstract
Almost all noise-robust ASR techniques discussed so far in the book have assumed the use of a single microphone device that captures distorted speech signals. We devote this chapter to the techniques developed with multiple devices. Due to the availability of cheap hardware, an increasing portion of devices features multiple sound capturing channels. This adds the spatial dimension to the otherwise only spectro-temporal processing of single-microphone systems. If the target speech and the interferers are spatially separated, beamforming and other multi-channel processing can greatly improve the target signal-to-noise ratio. It is particularly advantageous in the presence of nonstationary interferers ...