Augmentation System Design And Implementation Of Dsp-based Real-time Voice,

In speech communication speakers often interfered by background noise. This is particular the case in some special field such as proving ground with intense background noise, the local of military or naval maneuver and hand-free communication system in the car. The intense background noise distorts the speech and words are hardly intelligible. In order to improve the performance of speech communication and reduce the noise corrupted in speech a noise reduction or an also called speech enhancement is applied.Among of many algorithms spectral subtraction method has been one of the most well known techniques for noise reduction because it is simple and effective. The enhanced speech would be corrupted by a residual noise commonly called musical noise and unsuppressible background noise with this method. A lot of methods appears to modify the spectral subtraction method and make a great progress on suppressing musical noise, background noise and reducing speech distortion. But the performance of these methods got worse when the signal noise ratio is low.This thesis propose a modified spectral subtraction approach that allows better and more suppression of the noise and improve the intelligence of enhanced speech in a perceptual sense with perceptual weighting filter. Then a real-time speech enhancement system is implemented on TMS320VC5410 DSP.The results of experiments show that this system can suppress the noisy speech effectively on time with a low level of musical noise even at a low SNR.