Voice conversion for enhancing various types of body‐conducted speech detected with non‐audible murmur microphone.

Abstract

Our proposed statistical voice conversion approach to enhancing various types of body‐conducted speech detected with Non‐Audible Murmur (NAM) microphone is presented in this talk. NAM microphone, one of the body‐conductive microphones [Nakajima et al., IEICE Trans. Inf. and Syst., E89‐D, 1–8 (2006)], enables us to detect various types of body‐conducted speech such as extremely soft whisper, normal speech, and so on. Moreover, it is robust against external noise due to its noise‐proof structure. To make speech communication more universal by effectively using these properties of NAM microphone, body‐conducted speech enhancement techniques have been developed with a state‐of‐the‐art statistical voice conversion algorithm [Toda et al., IEEE Trans. ASLP, 15, 2222–2235 (2007)]. The proposed techniques would bring a new paradigm to human‐to‐human speech communication: e.g., the use of body‐conducted voiced speech for noise robust speech communication, the use of body‐conducted unvoiced speech for silent speech ...