Nonlinear Information Fusion in Multi-Sensor Processing – Extracting and Exploiting Hidden Dynamics of Speech Captured by a Bone-Conductive Microphone

  • Li Deng ,
  • Zicheng Liu ,
  • Zhengyou Zhang ,
  • Alex Acero

Proc. of the IEEE Workshop on Multimedia Signal Processing |

Published by Institute of Electrical and Electronics Engineers, Inc.

One well-known difficulty in creating effective human-machine interface via the speech input is the adverse effects of concurrent acoustic noise. To overcome this challenge, we have developed a joint hardware and software solution. A novel bone-conductive microphone is integrated with a regular air-conductive one in a single headset. These two simultaneous sensors capture distinct signal properties in the speech embedded in acoustic noise. The focus of this paper is exploration of the type of dynamic properties that are relatively invariant between the bone-conductive sensor’s signal and the clean speech signal; the latter would not be available to the recognizer. Our approach is based on a nonlinear processing technique that estimates the unobserved (hidden) vocal tract resonances, as a representation of such invariant hidden dynamics, from the available bone-sensor signal. The information about these dynamic aspects of the clean speech is then fused with other noisy measurements to aim at improving the recognition system’s robustness to acoustic distortion. The fusion technique is based on a combination of three sets of signals including the synthesized speech signal using the vocal tract resonance dynamics extracted nonlinearly from the bone-sensor signal.