Elderly speech collection for speech recognition based on crowd sourcing

Abstract

In this paper, a preliminary experiment on elderly speech data collection based on a crowd-sourcing approach is described. From this experiment, a usability study of the web platform used for this collection was conducted and a few changes were made in order to meet senior users' needs. The main goal of this work is to collect as much speech as possible from senior citizens in order to improve the current speech recognition accuracy rates, which does not perform well with speech uttered by the elderly. This work is under the scope of a larger project(1) focusing on the development of technologies and services to support healthy, productive and active citizens, using multimodal interfaces and particularly embedded speech technology.