Acoustic classification and automatic border of phrase and stress recognition in Polish

• Autorzy: Demenko G.
• Języki publikacji: EN

Abstrakty

EN

A classification of perceptively and linguistically verified intonation structures was made basing on acoustic qualities of speech signals. The importance of the eight selected parameters describing the set of elements under investigation was tested by means of statistical methods. Findings of discriminant analysis proved the possibility of a correct classification of unstressed vowels (in the range of 86-92%). For the remaining vowels (stressed and located at the end of a phrase) a significant degree of correct classification was obtained in the range of 54-73%. Far better classification results (88% at the average) were obtained by using a neural network. Identification of accents and phrase boundaries originating from the new data set showed the necessity to extend the linguistic base and repeat the network training.

• Wydawca: Instytut Podstawowych Problemów Techniki PAN ; Komitet Akustyki PAN ; Polskie Towarzystwo Akustyczne
• Czasopismo: Archives of Acoustics
• Rocznik: 1998
• Tom: Vol. 23, no. 2
• Strony: 159--178
• Opis fizyczny: Bibliogr. 16 poz., rys., tab., wykr.

Twórcy

autor

Demenko G.

• Institute of Linguistics, Adam Mickiewicz University, Poznań

Bibliografia

• [1] N. Cambell, Automatic detection of prosodic boundary in speech, Speech Communication, 13, 345-354 (1993).
• [2] G. Demenko, W. Jassem, M. Krzysko, Classification of basic FO patterns using discriminant functions, Phonetica, 41 (1988).
• [3] G. Demenko, L. Richter, P. Serwat, Analiza statystyczna wyników percepcyjnej oceny granic frazowych i akcentu w języku polskim, Materiały XLIII Seminarium z Akustyki, 167-172, (1996).
• [4] G. Demenko, Wyznaczniki fonetyczno-akustyczne granicy frazy i akcentu w języku polskim, Technologia Mowy i Języka, vol. 1, 101-124, Wrocław 1997.
• [5] P. Dumouchel, D. O’Shaughnessy, Prosody and continuous speech recognition, Proceedings of the 3rd European Conference on Speech and Technology, Eurospeech93, 2195-2198, Berlin 1993.
• [6] L. Frąckowiak-Richter, The duration of Polish vowels, Speech Analysis and Synthesis, v.3, 87-115, PWN, Warszawa 1973.
• [7] K. Hirose, A. Sakura, H. Konno, Use of prosodic features in the recognition of continuous Speech, Proceedings of ICSLP44, Yokohama, 1123-1126, (1994).
• [8] J. Imioloczyk, I. Nowak, G. Demenko, Implementacja systemu syntezy ciągłej mowy polskiej z tekstu ortograficznego wprowadzonego z klawiatury komputera typu PC, Prace IPPT , Warszawa, 11, (1993).
• [9] W. Jassem, Akcent języka polskiego, PAN, Wrocław 1962.
• [10] W. Jassem, G. Demenko, Fonetyczno-gramatyczna spójność frazy, Technologia mowy i języka, vol.l , 125-139, Wrocław 1997. [11] D. Scott, Duration as a cue to the perception of a phrase boundary, Journal of Acoustical Society of America, 71, 996-1007 (1982).
• [12] M. Steffen-Batóg, Analiza struktury przebiegu melodii polskiego języka ogólnego, Rozprawa doktorska, Poznań 1963.
• [13] R. Tadeusiewicz, Sieci neuronowe, Warszawa 1993.
• [14] C. Traber, Data-driven prosody generation using automatic learning procedures, Recueil des Publications et Communications Externes du Departement RCP, CNET, 159-162 (1997).
• [15] J. Vaissiere, The use of prosodic parameters in automatic speech recogniton, [in:] Recent Advances in Speech Understanding and Dialog Systems, H. Niemann, M. Lang, G. Sagerer [Eds.], Springer Verlag, 46, 71-99 (1988).
• [16] J.M. Zurada, Introduction to artifical neural systems, West Publishing Company, 1992.