Exploring Capabilities of Monolingual Audio Transformers using Large Datasets in Automatic Speech Recognition of Czech

dc.contributor.authorLehečka, Jan
dc.contributor.authorŠvec, Jan
dc.contributor.authorPražák, Aleš
dc.contributor.authorPsutka, Josef
dc.date.accessioned2023-01-30T11:00:27Z
dc.date.available2023-01-30T11:00:27Z
dc.date.issued2022
dc.description.abstract-translatedIn this paper, we present our progress in pretraining Czech monolingual audio transformers from a large dataset containing more than 80 thousand hours of unlabeled speech, and subsequently fine-tuning the model on automatic speech recognition tasks using a combination of in-domain data and almost 6 thousand hours of out-of-domain transcribed speech. We are presenting a large palette of experiments with various fine-tuning setups evaluated on two public datasets (CommonVoice and VoxPopuli) and one extremely challenging dataset from the MALACH project. Our results show that monolingual Wav2Vec 2.0 models are robust ASR systems, which can take advantage of large labeled and unlabeled datasets and successfully compete with state-of-the-art LVCSR systems. Moreover, Wav2Vec models proved to be good zero-shot learners when no training data are available for the target ASR task.en
dc.format5 s.cs
dc.format.mimetypeapplication/pdf
dc.identifier.citationLEHEČKA, J. ŠVEC, J. PRAŽÁK, A. PSUTKA, J. Exploring Capabilities of Monolingual Audio Transformers using Large Datasets in Automatic Speech Recognition of Czech. In Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH. New York: Red Hook, 2022. s. 1831-1835. ISBN: neuvedeno , ISSN: 2308-457Xcs
dc.identifier.doi10.21437/Interspeech.2022-10439
dc.identifier.isbnneuvedeno
dc.identifier.issn2308-457X
dc.identifier.obd43936705
dc.identifier.uri2-s2.0-85139048808
dc.identifier.urihttp://hdl.handle.net/11025/51163
dc.language.isoenen
dc.project.ID90140/Velká výzkumná infrastruktura_(J) - e-INFRA CZcs
dc.project.IDGA22-27800S/Využití vícemodálních Transformerů pro přirozenější hlasový dialogcs
dc.project.IDEF17_048/0007267/InteCom: VaV inteligentních komponent pokročilých technologií pro plzeňskou metropolitní oblastcs
dc.publisherInternational Speech Communication Associationen
dc.relation.ispartofseriesProceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECHen
dc.rightsPlný text není přístupný.cs
dc.rights© 2022 ISCAen
dc.rights.accessclosedAccessen
dc.subject.translatedspeech recognition, audio transformers, Wav2Vecen
dc.titleExploring Capabilities of Monolingual Audio Transformers using Large Datasets in Automatic Speech Recognition of Czechen
dc.typekonferenční příspěvekcs
dc.typeConferenceObjecten
dc.type.statusPeer-revieweden
dc.type.versionpublishedVersionen

Files

Original bundle
Showing 1 - 1 out of 1 results
No Thumbnail Available
Name:
Lehecka_Svec_Prazak_PsutkaJV-Exploring_Capabilties_Interspeech_2022.pdf
Size:
197.58 KB
Format:
Adobe Portable Document Format