Voice controlled wheelchairs: fine control by humming

Abstract

People without disabilities seamlessly control devices with their hands. Interestingly, their hands can perform coarse and fine control. Implementing smooth control for computerized systems is not straightforward and most of the time it is not intuitive either. Here we offer a solution to that problem: smooth control through humming. Voice commands have become ubiquitous in modern technology. Speech-to-text applications abound. Smooth control, on the other hand, has not been tackled yet. Here we design and implement a humming control technique, and demonstrate a hardware implementation with a powered wheelchair. Once actuated, the speed with which the chair moves will depend on the subtle variation on the fundamental frequency of the user's humming, acquired through an accelerometer measuring vocal cord vibration. We also discuss two signal processing techniques that handle commonly encountered issues when trying to resolve frequencies in real time data. The hardware implementation shows performance of 80% and higher in speech recognition for signal-to-noise ratio (SNR) higher than 8dB and 100% in smooth control and frequency detection for all tested SNRs. We also discuss potential applications of smooth humming control to other assistive technology.

Keywords: Accelerometry; Assistive technology; Humming; Power wheelchair; Smooth control; Voice control.