Optimization of Formant Feature Based Speech Recognition

摘要

The paper deals with the use of formant features in dynamic time warping based speech recognition. These features can be simply visualized and give a new insight into understanding the reasons of speech recognition errors. The formant feature extraction method, based on the singular prediction polynomials, has been applied in recognition of isolated words. However, the speech recognition performance depends on the order of singular prediction polynomials, whether symmetric or antisymmetric singular prediction polynomials are used for recognition and as well on the fact even or odd order of these polynomials is chosen. Also, it is important to know how informative separate formants are, how the speech recognition results depend on other parameters of the recognition system such as: analysis frame length, number of the formants used in recognition, frequency scale used for representation of formant features, and the preemphasis filter parameters. Properly choosing the processing parameters, it is possible to optimize the speech recognition performance. The aim of our current investigation is to optimize formant feature based isolated word recognition performance by varying processing parameters of the recognition system as well as to find improvements of the recognition system which could make it more robust to white noise. The optimization experiments were carried out using speech records of 111 Lithuanian words. The speech signals were recorded in the conventional room environment (SNR = 30 dB). Then the white noise was generated at a predefined level (65 dB, 60 dB and 55 dB) and added to the test utterances. The recognition performance was evaluated at various noise levels. The optimization experiments allowed us to improve considerably the performance of the formant feature based speech recognition system and made the system more robust to white noise.%Darbe yra nagrin?jamas formantini? po?ymi? naudojimas dinaminiu laiko skal?s kraipymu gr?stame kalbos atpa?inime. ?ie po?ymiai yra lengvai vizualizuojami ir suteikia nauj? ?ini? stengiantis suvokti kalbos atpa?inimo klaid? prie?astis. Izoliuot? ?od?i? atpa?inimui buvo pritaikytas i?sigimusiais prognoz?s polinomais gr?stas formantini? po?ymi? i?skyrimo metodas. Ta?iau, kalbos atpa?inimo tikslumas priklauso nuo to, ar atpa?inimui naudojami i?sigim? prognoz?s polinomai yra simetriniai ar antisimetriniai ir ar jie yra lygin?s ar nelygin?s eil?s. Taip pat svarbu i?tirti, kiek informatyvios yra atskiros formant?s, kaip kalbos atpa?inimo rezultatai priklauso nuo kit? atpa?inimo sistemos parametr?: analiz?s kadro ilgio, atpa?inimui naudojam? forman?i? skai?iaus, formantini? po?ymi? atvaizdavimui naudojamos da?ni? skal?s, pradin?s filtracijos filtro parametr?. Tinkamai parinkus ?iuos atpa?inimo sistemos parametrus galima pagerinti sistemos atpa?inimo tikslum?. ?io tyrimo tikslas yra kei?iant atpa?inimo sistemos parametrus optimizuoti formantiniais po?ymiais gr?sto izoliuot? ?od?i? atpa?inimo tikslum? ir surasti atpa?inimo sistemos patobulinimo būdus, kurie padidint? sistemos atsparum? balto triuk?mo poveikiui. Optimizavimo eksperimentai buvo atliekami naudojant 111 lietuvi? kalbos ?od?i? kalbos signalo ?ra?us. Kalbos signalai buvo ?ra?yti ?prastin?je kambario aplinkoje esant 30 dB signalo/triuk?mo santykiui. Tada buvo generuojamas 65 dB, 60 dB ir 55 dB baltas triuk?mas ir pridedamas prie testini? i?tarim? balso ?ra??. Atpa?inimo tikslumas buvo vertinamas esant ?vairiems u?triuk?minimo lygiams. Atlikti optimizavimo eksperimentai pad?jo ?ymiai pagerinti formantiniais po?ymiais gr?stos kalbos atpa?inimo sistemos tikslum? ir padaryti j? atsparesne balto triuk?mo poveikiui.