Název:	CNN-TDNN-Based Architecture for Speech Recognition Using Grapheme Models in Bilingual Czech-Slovak Task
Autoři:	Psutka, Josef Švec, Jan Pražák, Aleš
Citace zdrojového dokumentu:	PSUTKA, J. ŠVEC, J. PRAŽÁK, A. CNN-TDNN-Based Architecture for Speech Recognition Using Grapheme Models in Bilingual Czech-Slovak Task. In Text, Speech, and Dialogue 24th International Conference, TSD 2021, Olomouc, Czech Republic, September 6–9, 2021, Proceedings. Cham: Springer International Publishing, 2021. s. 523-533. ISBN: 978-3-030-83526-2 , ISSN: 0302-9743
Datum vydání:	2021
Nakladatel:	Springer International Publishing
Typ dokumentu:	konferenční příspěvek ConferenceObject
URI:	2-s2.0-85115207848 http://hdl.handle.net/11025/47248
ISBN:	978-3-030-83526-2
ISSN:	0302-9743
Klíčová slova v dalším jazyce:	Speech recognition;Multilingual training;Robustness;Acoustic modeling
Abstrakt v dalším jazyce:	Czech and Slovak languages are very similar, not only in writing but also in phonetic form. This work aims to find a suitable combination of these two languages concerning better recognition results. We would like to show such a contribution on the Malach project. The Malach speech of Holocaust survivors is highly emotional, filled with many disfluencies, heavy accents, age-related coarticulation, and many non-speech events. Due to the nature of the corpus, it is very difficult to find other appropriate data for acoustic modeling, so such a combination can significantly improve the amount of training data. We will discuss the differences between the phoneme and grapheme way of combining Czech with Slovak. We will also compare different architectures of deep neural networks (TDNN, TDNNF, CNN-TDNNF) and tune the optimal topology. The proposed bilingual ASR approach provides a slight improvement over monolingual ASR systems, not only at the phoneme level but also at the grapheme.
Práva:	Plný text je přístupný v rámci univerzity přihlášeným uživatelům. © Springer
Vyskytuje se v kolekcích:	Konferenční příspěvky / Conference Papers (KKY) OBD

Všechny záznamy v DSpace jsou chráněny autorskými právy, všechna práva vyhrazena.