Various services providing vocal answers related to telephone directory inquiries, weather forecast, mail order, etc., are now available to PSTN users using vocal servers. As the speech messages are produced by machines, they may suffer from some impairment.
In this Recommendation a method is defined for subjective performance assessment of the quality of speech of voice output devices. This method allows the comparison of several systems between them. It will be useful for system designers and service providers for checking the quality of their products.
This method is of the listening test type. Messages are presented aurally to subjects. The subjects express their opinion on one or more rating scales after having answered specific questions on the information contained in the messages. The results are measures of the perceived quality in several aspects, which makes it possible to compare the effectiveness of different speech synthesis systems.