Deep Learning-based Speech Recognition Algorithm for a Bionic Arm

This study concentrates on developing a deep-learning-based speech recognition algorithm that can be deployed on a bionic arm in the local language for an upper arm amputee. The speech recognition system ensures optimum performance with predefined actions based on the commands. The targeted operations to be done with the bionic arm are holding, gripping, giving, and fetching. CNN was used to develop a machine learning-based speech recognition model and it was incorporated into the bionic arm. For this, the Arduino nano ble 33 speech recognition module is used to control the arm based on the command given. The same was extended with coding and assembly modifications. In response, the arm responded to its predefined actions for the commands it has been programmed with.