LSTM-Based Speech Segmentation Trained on Different Foreign Languages

dc.contributor.authorHanzlíček, Zdeněk
dc.contributor.authorVít, Jakub
dc.date.accessioned2021-03-29T10:00:17Z
dc.date.available2021-03-29T10:00:17Z
dc.date.issued2020
dc.description.abstract-translatedThis paper describes experiments on speech segmentation by using bidirectional LSTM neural networks. The networks were trained on various languages (English, German, Russian and Czech), segmentation experiments were performed on 4 Czech professional voices. To be able to use various combinations of foreign languages, we defined a reduced phonetic alphabet based on IPA notation. It consists of 26 phones, all included in all languages. To increase the segmentation accuracy, we applied an iterative procedure based on detection of improperly segmented data and retraining of the network. Experiments confirmed the convergence of the procedure. A comparison with a reference HMM-based segmentation with additional manual corrections was performed.en
dc.format9 s.cs
dc.format.mimetypeapplication/pdf
dc.identifier.citationHANZLÍČEK, Z. VÍT, J. LSTM-Based Speech Segmentation Trained on Different Foreign Languages. In: Text, Speech, and Dialogue 23rd International Conference, TSD 2020, Brno, Czech Republic, September 8-11, 2020, Proceedings. Cham: Springer Nature Switzerland AG, 2020. s. 456-464. ISBN 978-3-030-58322-4, ISSN 0302-9743.cs
dc.identifier.doi10.1007/978-3-030-58323-1_49
dc.identifier.isbn978-3-030-58322-4
dc.identifier.issn0302-9743
dc.identifier.obd43930257
dc.identifier.uri2-s2.0-85091145791
dc.identifier.urihttp://hdl.handle.net/11025/43117
dc.language.isoenen
dc.project.IDGA19-19324S/Plně trénovatelná syntéza české řeči z textu s využitím hlubokých neuronových sítícs
dc.project.IDSGS-2019-027/Inteligentní metody strojového vnímání a porozumění 4cs
dc.project.IDLM2018140/E-infrastruktura CZcs
dc.publisherSpringer Nature Switzerland AGen
dc.relation.ispartofseriesText, Speech, and Dialogue 23rd International Conference, TSD 2020, Brno, Czech Republic, September 8-11, 2020, Proceedingsen
dc.rightsPlný text není přístupný.cs
dc.rights© Springeren
dc.rights.accessclosedAccessen
dc.subject.translatedSpeech segmentationen
dc.subject.translatedNeural networksen
dc.subject.translatedLSTMen
dc.titleLSTM-Based Speech Segmentation Trained on Different Foreign Languagesen
dc.typekonferenční příspěvekcs
dc.typeconferenceObjecten
dc.type.statusPeer-revieweden
dc.type.versionpublishedVersionen

Files