Is text-to-speech synthesis ready for use in computer-assisted language learning?

In Computer-Assisted Language Learning (CALL) environments, learners have been provided with a new way of spoken language input. A program, called Text-to-speech (TTS) synthesizer, is used to generate speech out of text input in the computer. This program can is beneficial for creation and editing of speech models, generation of speech models and feedback on demand. In addition, it can be used in talking dictionaries, talking texts, dictation, pronunciation training, dialogue partner.. etc. Many attempts were conducted to evaluate Text-to-speech (TTS) synthesis. They only addressed the comprehensibility of TTS synthesis, which make the program not widely accepted and used. In Computer-Assisted Language Learning (CALL) environments, and beside the focus on comprehensibility, the naturalness, accuracy, register and expressiveness of the output of TTS synthesis were main aspects of demands. This paper evaluates the quality of the output of four state-of-the-art French TTS synthesis systems, considering their use in the three different roles that TTS synthesis systems may assume within CALL applications, namely: reading machine, pronunciation model and conversational partner.

The results of this evaluation suggest that the best TTS synthesis systems are ready for use in applications in which they ‘add value’ to CALL, i.e. exploit the unique capacity of TTS synthesis to generate speech models on demand (An example of such an application is a dialogue partner). Unfortunately, most of the French systems cater attention to these 'add value' applications. The results of this study suggest that further research into the requirements imposed on the flexibility of TTS synthesis is necessary to draw any general conclusions about the readiness of TTS synthesis for use in CALL.

Regarding evaluation criteria, this study highlights the importance to CALL of two characteristics which are not currently addressed by general purpose evaluation tools,
namely naturalness and expressiveness. It is recommended that general purpose tools for the evaluation of TTS synthesis and comparative evaluation campaigns address these criteria.



* Reference: http://www.sciencedirect.com/science?_ob=MImg&_imagekey=B6V1C-4V761Y1-1-5&_cdi=5671&_user=912155&_orig=search&_coverDate=12%2F25%2F2008&_sk=999999999&view=c&wchp=dGLzVtz-zSkWz&_valck=1&md5=7acfb7c76f4794eb56c91beda3c8dcd4&ie=/sdarticle.pdf