Speech processing apparatus and mobile transceiver

摘要

The terminal has an estimator continuously forming estimated values (SNR(i)) of the signal-to-noise ratio of the speech signal. Power values, Px(i), of sample values of the speech signal (x(i)) are derived and smoothed. The minimus of a group of L successive smoothed power values is derived. The groups follow on successively without any gap. Each group contains at least as many smoothed power values as needed so that all power values belonging to any phoneme of the speech signal can be detected from any group. An actual value of the signal to noise ratio is then formed from the smoothed power values and the derived minimus.