Speech Enhancement for Laser Doppler Vibrometer Dealing with Unknown Irradiated Objects

摘要

The acquisition of distant sound has always been a hot researchtopic. Since sound is caused by vibration, one of the best methods formeasuring distant sound is to use a laser Doppler vibrometer (LDV). Thislaser has high directivity, that enables it to acquire sound from far away,which is of great practical use for disaster relief and other situations. However,due to the vibration characteristics of the irradiated object itself andthe reflectivity of its surface (or other reasons), the acquired sound is oftenlacking frequency components in certain frequency bands and is mixed withobvious noise. Therefore, when using LDV to acquire distant speech, if wewant to recognize the actual content of the speech, it is necessary to enhancethe acquired speech signal in some way. Conventional speech enhancementmethods are not generally applicable due to the various types of degradationin observed speech. Moreover, while several speech enhancement methodsfor LDV have been proposed, they are only effective when the irradiatedobject is known. In this paper, we present a speech enhancement method forLDV that can deal with unknown irradiated objects. The proposed methodis composed of noise reduction, pitch detection, power spectrum envelopeestimation, power spectrum reconstruction, and phase estimation. Experimentalresults demonstrate the effectiveness of our method for enhancingthe acquired speech with unknown irradiated objects.