Speech generation in a spoken dialogue system

[摘要] ENGLISH ABSTRACT: Spoken dialogue systems accessed over the telephone network are rapidly becoming morepopular as a means to reduce call-centre costs and improve customer experience. It isnow technologically feasible to delegate repetitive and relatively simple tasks conductedin most telephone calls to automatic systems. Such a system uses speech recognition totake input from users. This work focuses on the speech generation component that aspecific prototype system uses to convey audible speech output back to the user.Many commercial systems contain general text-to-speech synthesisers. Text-to-speechsynthesis is a very active branch of speech processing. It aims to build machines thatread text aloud. In some languages this has been a reality for almost two decades. Whilethese synthesisers are often very understandable, they almost never sound natural. Theoutput quality of synthetic speech is considered to be a very important factor in the user'sperception of the quality and usability of spoken dialogue systems.The static nature of the spoken dialogue system is exploited to produce a customspeech synthesis component that provides very high quality output speech for the particularapplication. To this end the current state of the art in speech synthesis is surveyedand summarised. A unit-selection synthesiser is produced that functions in Afrikaans,English and Xhosa.The unit-selection synthesiser selects short waveforms from a recorded speech corpus,and concatenates them to produce the required utterances. Techniques are developed fordesigning a compact corpus and processing it to produce a unit-selection database. Speechmodification methods were researched to build a framework for natural-sounding speechconcatenation. This framework also provides pitch and duration modification capabilitiesthat will enable research in languages such as Afrikaans and Xhosa where text-to-speechcapabilities are relatively immature.

[发布机构] Stellenbosch University