Improving Bone-Conducted Speech Quality via Neural Network
Tetsuya Shimamura, Jun'ichiro Mamiya, Toshiki Tamiya · 2006
The quality of bone-conducted speech is low, but bone-conducted speech itself is not affected by noise. In this paper, we take into account such properties of bone-conducted speech, and derive a reconstruction filter to improve the quality of bone-conducted speech. The reconstruction filter is designed by learning a neural network on the basis of the bone-conducted speech and normal speech obtained from a speaker. Experimental results show that the reconstructed speech signal has better quality than the bone-conducted speech signal