Massachusetts Science and Engineering Fair (MSEF) Project

Title: Machine Learning based Speech Recognition App for individuals with speech disorders
Problem and Societal Impact: 40 million Americans and over 300 million people worldwide have speech disorders. Even though Speech Recognition is becoming ubiquitous, there are no good solutions available for people with speech disorders.
Goal: Train a Speech Recognition model to recognize words for a target user using voice clips from that user. Once the model is created, it should recognize the words spoken by the same user on an Android app with over 60% accuracy.
Software:
- Tensorflow: An open-source machine learning platform by Google; allows to build neural networks with many layers.
- Anaconda: Environment to develop and test the code in
- Python: Programming language used for data preparation of customer model
- Android Studio: Development environment for Android
Research Paper
Video that was submitted for the Science Fair