This thesis investigates using piezoelectric pressure sensors in a novel wrist-worn device that recognizes finger gestures using signals recorded at the wrist. The device measures tendon shape changes and vibration signals, which pass through an analog signal conditioning circuit connected to a microcontroller. We designed gesture recognition software that uses a sliding window function to detect the timing of a gesture. Our recognition software investigated three machine learning classifiers—nearest-neighbour, decision tree, and support vector machine—trained on features of the piezoelectric signal. We designed an experiment to test the recognition system on three types of individual finger tap gestures. The highest mean gesture detection accuracy was 93% across all ten participants. For classifiers trained per participant, the highest mean accuracies for finger and gesture type classification were 93% and 98%, respectively. The device shows promise as a wearable computer interface, which could enable mobile text input via air typing.