Development and Comparative Analysis of a Novel Neural Model for Marathi Alphabet Recognition for Hearing Aid Users
- Prashant G. Patil
- Arun K. Mittra
- Vijay S. Chourasia
Frequency Compression, Frequency Transposition, Neural Network, Hearing Aid, Hearing Loss, Hearing Aid users.
In this Paper weassociates differences between high frequency hearing loss reduction methods which is useful in improvement in speech intelligibility of HA users. Frequency Compression (M1) and Frequency transposition (M2) was usuallyimplemented by many researchers &accepted by numerousHA users. In certain circumstances both methods was not acceptable by HA users. We proposed two feed forward back propagation neural network based Frequency Compression (M3) & Frequency Transposition Methods (M4) methods. Four algorithm was designed using MATLAB & tested on Laptop with wired ear set. The performance parameter of all four methods was found by using audiogram of all 6 HA users. These Parameters are fitted to corresponding MATLAB based algorithm. Regional Marathi language spoken HA users are selected as participants for recognition & Performance evaluation test of all four algorithm. It is observed that average recognition rate by group of 6 HA user increased for vowels & consonants from minimum 2 % to 14 %.All these tests are considered insensitive for objectively and accurately measuring aspects oflisteners' speech perception abilities as a reflection of their performance in realistic listening situations.As a result of this Experimentation, the participant will be able to compare the average performance of all four methods and determine if his or her performance as expected on various processing methods for improvement in high frequency region.
Prashant G. Patil, Arun K. Mittra, Vijay S. Chourasia. "Development and Comparative Analysis of a Novel Neural Model for Marathi Alphabet Recognition for Hearing Aid Users".INTERNATIONAL JOURNAL OF ENGINEERING DEVELOPMENT AND RESEARCH ISSN:2321-9939, Vol.5, Issue 4, pp.649-656, URL :https://rjwave.org/ijedr/papers/IJEDR1704104.pdf
Volume 5 Issue 4
Pages. 649-656