Synthetic speech optimised for low cognitive load

  • Position identifier: ESR6
  • Host partner: UEDIN

Objectives

Synthetic speech (generated by computer from text input) has dramatically improved in naturalness and intelligibility in recent years. Yet, in adverse conditions such as noisy environments, it is still much harder to understand than natural speech. Even in quiet conditions, we believe that synthetic speech requires more effort from listeners: it is widely-acknowledged to be ‘hard to listen to’. This probably makes it more distracting than natural speech in many situations, such as in-car navigation.

The objectives of this project are to:

  • Investigate the cognitive load imposed by state-of-the art synthetic speech in both quiet and noise, using the multi-instrument techniques provided by other researchers in the project;
  • Discover which aspects of synthetic speech contribute most to the ‘hard to listen to’ effect (which is hypothesised to be an increased cognitive load);
  • Design, implement and test novel forms of synthetic speech that impose the lowest possible cognitive load on listeners.

We expect that, by solving these problems, we will also make advances in the quality and naturalness of synthetic speech.

The project will be supervised by Prof. Simon King and will be carried our in collaboration with partners Tobii Dynavox (Sweden), Radboud University Nijmegen (Netherlands) and the University of Crete (Greece).