Voice Recognition Research

  • Goals: Determine best and worst practices for voice recognition software, including accuracy and performance thresholds
  • Method: Measured and assessed seven types of voice recognition software via multiple rounds of usability testing, and evaluated how user perceptions are affected by degraded performance across various task commands and dictation
  • Results: Best and worst practices were identified, and user perceptions about acceptability of accuracy were determined

Images depicting the voice recognition software analyzed (left), and a chart showing the acceptability of task accuracy (right).