Impact of Technology in Filter Design for Noise Removal from Pathological Noisy Speech Signal & its Preprocessing Syed Mohammad Ali Department of Electronics & Telecommunication Engineering, Anjuman College of Engineering & Technology, Nagpur, India Email: ali_acet2003@yahoo.co.in Pradeep Tulshiram Karule Department of Electronics Engineering, Yeshwantrao chavan College of Engineering, Wanadongri, Nagpur, India Email: ptkarule@rediffmail.com Abstract—In the recent year the trend towards automated analysis of pathological noise signal has gain momentum. The awkwardness of analog equipment has simulated development of digital computer techniques for processing and analysis of pathological speech signal in patient care system. The above filter design techniques & prepossessing of speech signal can be used in any speech processing application. This paper discusses pathological speech signal of patients and their preprocessing. In prepossessing, Speech signal is passed through Moving Average (M.A) filter 1 , High pass (H.P) filter for removal of noise. The output of filter is framed & these frames are passed through window. Typically, hamming window is used. This preprocessed output can be used for pathological voice recognition, speech identification, speaker identification & many more application. Index Terms—Speech signal, moving average filter, highpass filter, framing, windowing I. INTRODUCTION A computerized technique for pathological voice recognition has received a greater attention from researchers in the last decade. Speech processing has proved to be excellent tool for speech disorder detection [1]. Pathological voice signal of patient from Dr. Naresh Agrawal Hospital, ENT Surgeon & Government Medical College & Hospital, Nagpur has been taken. The signals are recorded keeping mic two inch away from mouth using voice recorder of window XP. The sampling frequency is chosen to 11025 samples /sec. The patient has pronounced, ‘a’ which is vowel, then ‘ah’ which is vowel with consonant and a word ‘Hello’ for two sec, these signals are noisy & noise needs to be removed. So 1 Averaging of three samples gives best result. Manuscript received December 31, 2012; revised January 25, 2013; accepted February 15, 2013. filters are designed. The signal is passed through filter & then framing & windowing is done. The output of window is called prepossessed output, which can be used for further application like diagnosis of disease, speaker recognition, & speech recognition. Physicians often use invasive techniques like endoscopy to diagnose symptoms of vocal fold disorders however it is possible to diagnose disease using certain feature of speech signal [1]. Speech signal is sinusoidal signal having different frequency, different amplitude & different phase. It is given by the expression given below [2]. i i N 1 i i A (t)sin[2πF(t)t θ (t)]    (1) where, Ai (t), Fi (t) & i (t) are the sets of amplitudes, frequencies & phases respectively, of the sinusoids as shown in Fig. 1. Figure 1. A speech signal Voice & speech production requires close cooperation of numerous organs which from the phonetic point of view may be divided into organ.  Lungs, Bronchi, Tracheas (producing expiration air steam necessary for phonation)  Larynx (amplifying the initial tone)  Root of the tongue, throat, nasal cavity, oral cavity (forming tone quality & speech sound) [3]. Speech signal in non-intrusive in nature & it has potential for providing quantitative data with reasonable analysis time. So study of speech signal of pathological voice has become an important topic for research as it reduces work load in diagnoses of pathological voices [4].