Suppression of Internal Noise for Speech Recognition of Small Robots

Speech recognition using a robot has difficulties because the robot makes noise by itself. Especially, it is di冊cult fbr a small robot to reduce the internal noise by robot design. In this paper, two new methods are proposed that supresses internal noise of small robots. These methods are based on spectral subtraction(SS). The difference of the proposed methods from the orininal SS is that the proposed methods use the estimated noise spectrum depending on the motion of the robot. One method, called MDSS, prepares the noise spectrums for all motions. Another method, called NPSS, predicts the noise spectrum from angular velocities of all joints of the robot using a neural network. From the results of the comparison among the original SS and the proposed methods, the proposed methods outperformed the conventional SS. The MDSS method gave good results when the noise within one motion was stationary, while the NPSS worked well even when the noise of the motion was non-stationary.