Abstract: A system determines a power spectral density associated with an audio signal that includes a speech signal and/or a noise signal. The system updates an autocorrelation function of the audio signal from samples in the audio signal, estimates an autocorrelation function of the speech signal from the updated autocorrelation function of the audio signal, and calculates a power spectral density of the speech signal using the estimated autocorrelation function. The system then determines the power spectral density of the audio signal from the calculated power spectral density of the speech signal.

Abstract: A method for signal detection uses a likelihood ratio derived from the received signal to produce an estimate of a speech signal that has been corrupted by noise during transmission. The received signal is input to a receiver filter and a voice-activity detector. The receiver filter filters the received signal to produce a filter output signal. The voice-activity detector generates a likelihood ratio based on the received signal, which is then used to produce a speech-probability estimate indicating the probability that the received signal includes a speech signal. The filter output signal is combined with the speech-probability estimate output from the voice-activity detector to generate a soft estimate of the original speech signal.