An Assistive tool for Orthopedic Students: CNN-LSTM based Continuous Speech Recognition System for Writing Exams

In today’s fast-paced digital world, technology has become an indispensable tool for enhancing the quality of life for people with various disabilities. Individuals with Orthopedic challenges, such as limb disabilities or motor impairments, significantly impact a person’s ability to communicate effectively, hindering their participation in social and professional spheres. Orthopedic challenged people often face difficulties in writing, drawing, or expressing themselves in traditional mediums. These challenges can lead to frustration, isolation, and hindered educational or professional opportunities for them. Orthopedic students due to their impairment have difficulty in writing exams and sometimes they need human scribe to assist them. However, one promising technology for addressing their challenges is in the realm of speech technology. The main objective of this work focuses on developing a deep learning system that converts the dictated speech to text for the benefit of orthopedic students of India to continue their education. In this work, a CNN-LSTM based speech recognition system is developed and trained using the IndicTTS speech dataset. Achieving an overall accuracy of 74% for simple English sentences with 1,132 training samples, the results demonstrate the promising potential of using speech recognition as an assistive technology for writing exams by the orthopedic students.