Finnish parliament ASR corpus

dc.contributorAalto Universityen
dc.contributor.authorVirkkunen, Anjaen_US
dc.contributor.authorRouhe, Akuen_US
dc.contributor.authorPhan, Nhanen_US
dc.contributor.authorKurimo, Mikkoen_US
dc.contributor.departmentDepartment of Information and Communications Engineeringen_US
dc.descriptionFunding Information: This work has been supported by the MeMAD project of the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 780069) and the Academy of Finland project funding grants numbers 329267, 337073, and 345790. We also thank Aalto ScienceIT for providing us with computational resources. Publisher Copyright: © 2023, The Author(s). | openaire: EC/H2020/780069/EU//MeMAD
dc.description.abstractPublic sources like parliament meeting recordings and transcripts provide ever-growing material for the training and evaluation of automatic speech recognition (ASR) systems. In this paper, we publish and analyse the Finnish Parliament ASR Corpus, the most extensive publicly available collection of manually transcribed speech data for Finnish with over 3000 h of speech and 449 speakers for which it provides rich demographic metadata. This corpus builds on earlier initial work, and as a result the corpus has a natural split into two training subsets from two periods of time. Similarly, there are two official, corrected test sets covering different times, setting an ASR task with longitudinal distribution-shift characteristics. An official development set is also provided. We developed a complete Kaldi-based data preparation pipeline and ASR recipes for hidden Markov models (HMM), hybrid deep neural networks (HMM-DNN), and attention-based encoder-decoders (AED). For HMM-DNN systems, we provide results with time-delay neural networks (TDNN) as well as state-of-the-art wav2vec 2.0 pretrained acoustic models. We set benchmarks on the official test sets and multiple other recently used test sets. Both temporal corpus subsets are already large, and we observe that beyond their scale, HMM-TDNN ASR performance on the official test sets has reached a plateau. In contrast, other domains and larger wav2vec 2.0 models benefit from added data. The HMM-DNN and AED approaches are compared in a carefully matched equal data setting, with the HMM-DNN system consistently performing better. Finally, the variation of the ASR accuracy is compared between the speaker categories available in the parliament metadata to detect potential biases based on factors such as gender, age, and education.en
dc.description.versionPeer revieweden
dc.identifier.citationVirkkunen , A , Rouhe , A , Phan , N & Kurimo , M 2023 , ' Finnish parliament ASR corpus : Analysis, benchmarks and statistics ' , LANGUAGE RESOURCES AND EVALUATION , vol. 57 , no. 4 , pp. 1645-1670 .
dc.identifier.otherPURE UUID: a770fab9-0811-49e6-9f84-10de4f0f12c4en_US
dc.identifier.otherPURE ITEMURL:
dc.identifier.otherPURE LINK:
dc.identifier.otherPURE FILEURL:
dc.relation.ispartofseriesLANGUAGE RESOURCES AND EVALUATIONen
dc.subject.keywordParliament speech dataen_US
dc.subject.keywordSpeech recognitionen_US
dc.titleFinnish parliament ASR corpusen
dc.typeA1 Alkuperäisartikkeli tieteellisessä aikakauslehdessäfi