An Intelligent System for Conversion of Bangla Sign Language into Speech

Abstract

Bangla sign language conversion into speech is a key to help vocally impaired people to express their feelings easily. In this wok, we develop a pair of smart gloves with sensors and microcontroller that can convert Bangla sign language into Bangla speech. Flex sensors are used to measure the bending of fingers, accelerometer and gyroscope sensor are used to measure hand position and movements. Filter these data and send it to PC over Bluetooth. The system manipulate the value string and dynamically match it with database to find out the corresponding word's sound track which is play on speakers. On-board experiments demonstrate the effectiveness of the approach. The proposed system is evaluated with 8 people with 248 samples of 72 words. Experimental results show that the system is functioning well and can detect the sign language with 99.5% accuracy. The user can change the pronunciation of words with the variations of tense and pronoun in Bangla sentences.