Publication Details
Advances in Acoustic Modeling for the Recognition of Czech
Glembek Ondřej, Ing., Ph.D. (DCGM FIT BUT)
Karafiát Martin, Ing., Ph.D. (DCGM FIT BUT)
Automatic Speech Recognition, LVCSR system, acoustic modeling, HLDA, VTLN, CMLLR, lectures recognition
The paper is on Advances in Acoustic Modeling for the Recognition of Czech
This paper presents recent advances in Automatic Speech Recognition for the Czech Language. Improvements were achieved both in acoustic and language modeling. We mainly aim on the acoustic part of the issue. The results are presented in two contexts, the lecture recognition and SpeeCon+Temic test set. The paper shows the impact of using advanced modeling techniques such as HLDA, VTLN and CMLLR. On the lecture test set, we show that training acoustic models using word networks together with the pronunciation dictionary gives about 4-5% absolute performance improvement as opposed to using direct phonetic transcriptions. An effect of incorporating the "schwa" phoneme in the training phase shows a slight improvement.
@INPROCEEDINGS{FITPUB8724, author = "Ji\v{r}\'{i} Kopeck\'{y} and Ond\v{r}ej Glembek and Martin Karafi\'{a}t", title = "Advances in Acoustic Modeling for the Recognition of Czech", pages = "357--363", booktitle = "Proc. 11th International Conference on Text, Speech and Dialogue", series = "Lecture Notes in Computer Science", volume = 5246, year = 2008, location = "Berlin, DE", publisher = "Springer Verlag", ISBN = "978-3-540-87390-7", language = "english", url = "https://www.fit.vut.cz/research/publication/8724" }