Joint estimation of time delay and pitch of voiced speech signals

The paper shows that the time delay of a voiced speech signal arriving at different microphones in a microphone array can be more accurately estimated from low signal to noise ratio data if the signal's quasi-periodic property is effectively used. We propose a joint estimator of the time delay and the pitch period based on the maximum likelihood method. For a periodic signal in noise, the proposed estimator is shown analytically to use the data more efficiently than the correlation method, yielding a more robust performance. The simulation tests show that the new delay estimator achieves a higher accuracy than the correlation method for speech data with moderately low signal to noise ratio.