Reduce the Complexity of Estimation a 2D Horizontal Directional Sound Using Cross-correlation Method for the Recognition of Robot Voice Directions

Abstract

Developing robots ability to be able to act similar to human being most for being substituted from human being in the risky and dangerous work in the future. Perceiving sound source origin is important to help robots move toward to the origin of sound. In addition, this idea can be applied as the leading system for the hearing disability people. Therefore, this research proposes Cross-correlation method to estimate the direction of sound from two microphones to be presented as human begins left ear and right ear with the qualifications of the transfer function which is related to the head (Head-Related Transfer Function: HRTF) and reduces the complexity of the sound direction estimation with the application of polynomial regressions. To estimated the tendency of correlation of sound direction in real time is faster. The method that has presented shown the approximation error as root mean square only 0.96 degrees.