AN EMPIRICAL ANALYSIS OF STUTTERING AND STAMMERING SPEECH (SS-S) USING ERGONOMICS

A Lalitha Venkatesan

Abstract: This paper describes about the state of art of speech recognition which will be helpful in identifying the stuttering and stammering speech of a person (disability). Every speech that is spoken should have proper quality tone and utterance but stammering speech is exceptional, as it lacks clarity. Successful communication depends on intonation, stress, rhythm where those attributes are very less in SS speech. To overcome this issue, there are few speech therapy process which are carried out (medically) those are less responsive and long term training is needed and cost effective too. To deal with this problem an ergonomic approach used in form of Mental Speech Model (MSM) to enhance the speech quality of every utterance of word. To create a MSM, a deep learning technique is needed which involves with Deep Neural Networks (DNN) over the SS speech. This model will have two repositories, one speech of normal person and other speech of abnormal person (SS-S) which is compared and matched to reflect the quality of speech utterance. The ergonomic approach that is used here can also act as a Psycho-Linguistic Approach (PLA) to train the system to improvise the speech quality to overcome the difficulties with stuttering and stammering speech. The ergonomic approach or PLA is effective in terms of speech process and cost.

Keywords: stuttering, stammering, speech quality, DNN, MSM, PLA

DOI: https://doi.org/10.15623/ijret.2016.0519005