Noise enhancement method for speech signal processing / Muhamad Arif Hashim and Roshidi Din

Speech recognition is like biometric (eg. smartcard) that based on person voice (speech) which is convert to signal, recognize whether the intended person is genuine or not. Basically, signal consists two kind of information: message information and noise information and as well as speech signal, it contains speech information and noise information. Generally, noise information is not needed and causes trouble by disturbing the speech signal and adds errors to the signal. To overcome this problem, speech enhancement method is used to separate speech information and noise from speech signal using several kinds of techniques such as Kalman Filtering and Signal-to-Noise ratio (SNR) and enhance the intended feature (speech information). In this paper; our main intention on different method called noise enhancement method, which based on speech enhancement method to recognize the intended speech signal and learning noise behaviors and its usefulness. The relationship between noise and speech signal will be discuss including chronology of speech signal processing and types of noise that appears. We state steps on doing the noise enhancement method and some mathematical basis will be presented. Finally, we will discuss challenge that occur and conclude it based on based on the discussion. Hopefully, it will give a new direction of noise and speech's intention not just in information communication and technology field but also in speech research, sound engineering and other related fields.