Noise spectrum subtraction method and system

1. A method for reducing noise in a voice signal, the method comprising:

(a) processing a digital signal representative of the voice signal including a speech component and a noise component, said processing comprising applying linear prediction coding (LPC) analysis to said digital signal thereby obtaining a compressed digital signal representative of said voice signal; and

(b) processing the compressed digital signal for determining a power spectrum of the noise component, thereby enabling to subtract the noise component from the compressed digital signal.

2. The method according to claim 1 , wherein said compressed digital signal is based on a set of (LPC) coefficients and a residual signal, said processing comprising parameterization of the residual signal.

3. The method according to claim 2 , wherein the processing of the compressed digital signal comprises:

carrying out said determining of the power spectrum of the noise component of said compressed digital signal during a non-speech activity, and calculating its average value;

calculating a power spectrum estimator of the compressed digital signal with a reduced noise component;

determining an autocorrelation function of the compressed digital signal with the reduced noise component; and

determining a set of modified LPC coefficients from the autocorrelation function.

4. A method for processing a voice signal to reduce a noise therefrom, the method comprising:

(a) providing a digital signal representative of said voice signal including a speech component and a noise component;

(b) applying linear prediction coding (LPC) analysis to the digital signal, thereby obtaining a compressed digital signal representative of said voice signal, wherein said compressed digital signal is based on a set of LPC coefficients and a residual signal;

(c) determining a power spectrum of the noise component during a non-speech activity, and calculating its average value;

(d) calculating a power spectrum estimator of the compressed digital signal with reduced noise component;

(e) determining an autocorrelation function of the compressed digital signal with the reduced noise component; and

(f) determining modified LPC coefficients representing the speech component with reduced noise spectrum from the autocorrelation function.

5. A voice processing unit for use in a voice operated system, the voice processing unit comprising a noise reduction utility interconnected between a voice coding utility and a voice recognition utility, the voice coding utility being configured and operable to process a digital signal representative of an input voice signal, including a speech component and a noise component, by applying linear prediction coding (LPC) analysis to said digital signal thereby obtaining a compressed digital signal representative of said input voice signal, the noise reduction utility being configured and operable for receiving the compressed digital signal, processing it to determine a power spectrum of the noise component, and generating an output compressed digital signal with reduced noise spectrum.

6. A voice operated system comprising: an input port for receiving an input voice signal; an analog-to-digital converter for processing the input signal to generate a digital output indicative thereof; a voice processing utility for processing the digital signal by applying thereto linear prediction coding (LPC) analysis and generating a compressed digital signal, representative of the input voice signal, said compressed digital signal being in the form of a set of LPC coefficients and a residual signal; a voice processing unit; a system interface utility; and a control module, which is interconnected between the voice processing utility and the voice processing unit, and is connected to the system interface to operate it in response to a speech signal; the voice processing unit comprising:

a noise reduction utility coupled to the voice processing utility for processing said compressed digital signal to determine a power spectrum of the noise component, and generating an output compressed digital signal with reduced noise spectrum; and

a voice recognition utility coupled to the noise reduction utility for processing said output compressed digital signal with reduced noise spectrum.