This paper presents a new speech enhancement method for hands-free telephones in vehicles.This new method uses two microphones,one of them for input of the noisy speech signal.The second microphone is used for speech/siglence separation when estimating the noise power.The noise reduction filter is based on a modified power specitral subtraction where the 'a priori' signal-to-noise ratio (SNR) is the main parameter.the new method allows a good noise reduction without introducing the 'musical noise' in the processed speech.Also,the proposed method presents a low computational load and it can be implemented for real-time operations.
展开▼