SPEECH QUALITY UNDER HEAVY NOISE CONDITIONS IN HANDS-FREE COMMUNICATION

摘要

When the noise in an audio signal made up of both speech and noise is suppressed, the quality of the speech in the audio signal is usually degraded. The speech obtained from a noise-suppressed audio signal is improved by determining linear predictive coding (LPC) characteristics of the audio signal without or prior to noise suppression and by determining the LPC characteristics of the noise-suppressed audio. The convolution of those differing characteristics provides an improved-quality speech signal, with the original noise level reduced or suppressed.

主权项

1. A method of improving speech quality in an audio signal comprising speech and noise, after noise in the audio signal is suppressed, the method comprising:
receiving a first audio signal comprising speech and noise; determining characteristics of the first audio signal and generating a linear predictive coding (LPC) representation of said first audio signal; providing the first audio signal to a noise suppressor, which is configured to suppress at least some of the noise in the first audio signal and to thereby produce a noise-reduced audio signal; generating an error signal using liner predictive coding (LPC) estimation method from the noise-reduced audio signal, the error signal comprising speech in the first audio signal after at least some of the noise in the first audio signal is removed; applying the error signal to the LPC representation of the first audio signal to synthesize a second audio signal having a reduced distortion speech and speech quality better than the speech in the first audio signal; and providing the second audio signal to an audio signal transducer configured to produce audible sound waves from the second audio signal.