Replacing a Human Agent by an Automatic Reverse Directory Service
Agents who answer the calls in a reverse directory service have to face a considerable challenge: they need to communicate proper names (such as the names of persons, companies and streets). Their pronunciation is frequently irregular and their spelling is not obvious. The authors developed a TTS specialized for this task, i.e. the reading of names and addresses in order to create an automatic reverse directory service. The novelty of our system compared to others developed earlier is that we employed a new reading mode and optimized the acoustic database based on an extensive analysis of Hungarian proper names. This resulted in high intelligibility and naturalness. Our system was launched as a service of T-Mobile Hungary. The specialized TTS can also be the basis of other applications in the future, such as location based services.
Unable to display preview. Download preview PDF.
- 1.Spiegel, M. F., “Coping with Telephone Directories that Were Never Intended for Synthesis Applications”, Proc. of ESCA-NATO/RSG 10 Tuto-rial and Workshop on Applications of Speech Technology, Lautrach, Germany, 1993, pp. 19-22Google Scholar
- 2.Belhoula, K., “A Concept for the Synthesis of Names”, Proc. of ESCA-NATO/RSG 10 Tutorial and Workshop on Applications of Speech Technology, Lautrach, Germany, 1993, pp. 167-170Google Scholar
- 3.Nebbia, L., Quazza, S., and Salza, P. L., “A Specialised Speech Synthesis Technique for Application to Automatic Reverse Directory Service”, Proc. of IVTTA ’98, IEEE-ESCA Workshop on Interactive Voice Tech-nology for Telecommunications Applications, Torino, Italy, 1998, pp. 223-228Google Scholar
- 4.Lundin, F. J., “The Swedish Automatic Reverse Directory Service”, Proc. of IVTTA ’98, IEEE-ESCA Workshop on Interactive Voice Technology for Telecommunications Applications, Torino, Italy, Sept, 1998, pp. 219-222Google Scholar
- 7.Olaszy, G. and Németh, G., “IVR for Banking and Residential Telephone Subscribers Using Stored Messages Combined with a New Number-to-Speech Synthesis Method”, in D. Gardner-Bonneau ed., Human Factors and Interactive Voice Response Systems, Kluwer, 1999, pp. 237-255Google Scholar
- 8.Fék M., Németh G., Olaszy G., Gordos G.: “Megértést segítő részletező gépi névfelolvasás magyar nyelvre” (Automatic syllabification support-ing understanding for Hungarian name-reading), Proc. of 2nd Hungarian Computational Linguistics Conference, Szeged, 2004, pp. 301-306Google Scholar