Development of Microcontroller based Speech Recognition Using CMSIS-NN Keyword Spotting

Ang, Siok Cheng (2021) Development of Microcontroller based Speech Recognition Using CMSIS-NN Keyword Spotting. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Ang Siok Cheng.pdf Restricted to Registered users only Download (2MB)

Abstract

Keyword Spotting is categorized under Speech Recognition which is a mechanism used by machine or any program to identify the word spoken by the user. This technology is believed to make human lives easier. Instead of pressing buttons to control or command a machine like UI (User Interface), speaking has become an option with speech recognition. Amazon Alexa, Apple’s Siri and Google’s Assistant are all successful examples of speech recognition that is used in our daily life. They were created to assist user to accomplish simple tasks like playing certain music, opening application and googling a word. Samsung Smart Fridge 2019 came up with a food-list-feature, which remind the user whenever a food in the list is about to expire. A minor flaw to this feature is the method proposed to input the food list, which is manually typed into the list. The solution proposed to address the problem is to replace the input-method with speech recognition. The final outcome of this project is to use an additional microcontroller-based development board to aid the input process. This project helps to increase the popularity of the food-list-feature, which helps avoid wasting food.