An AAC Application Using Speaking Partner Speech Recognition to Automatically Produce Contextually Relevant Utterances: Objective Results

Abstract

The purpose of this study was to evaluate the efficacy of a computer program, Converser, which uses the speech recognition of the speaking partner to predict contextually appropriate utterances and to store relevant noun phrases. The dyads (which consisted of a program user and a speaking partner) tested Converser in two communication tasks: a conversation and an interview. Two conditions were applied: A simple communication aid with an alphabet layout without Converser (alpha-only condition), and an identical aid with Converser (alpha-converser condition). Objective measurements were made concerning the program user's rate and Converser usage. The results showed that the alpha-converser condition generated a faster communication rate than the alpha-only condition. Also, the program users utilized the stored noun phrases in their productions more than the full utterances. Converser allowed for periodic instances of a fast rate of speech, which appeared to have a positive impact on communication.