TCNSpeech: A Community-Curated Speech Corpus for Sermons

Wuraola Fisayo Oyewusi, Sharon Ibejih, Soromfe Uzomah, Elizabeth Mawutin Joseph, Jon Cynthia, Folakunmi Ojemuyiwa, Benedicta Johnson-Onuigwe, Omolola Taiwo, Akintunde Akinpelumi, Olabisi Adesina, Ayodele Noutouglo, Adeola Adeleke Adeoba, Andrew Akoh, Chukwuemeka Nwachukwu, Opeyemi Agbabiaje, Itunu Falade, Olukemi Erhunmwunsee, Oluwatobiloba Dada, Olúwatóbi David OSIBELUWO, Ehis Akene et al. (12 additional authors not shown)

Keywords: ASR, Sermon, Speech To Text, Nigerian, TCNSpeech

TL;DR: A dataset of sermons curated by a community for Automatic Speech Recognition tasks

Abstract: In this work we present TCNSpeech, a community-curated multispeaker sermon corpus for speech recognition tasks. It contains a total of 24 hours of English audio data recording, chunked and transcribed. The context of the dataset is domain-specific for sermons in Nigerian English accent and a use case for community data curation. The dataset will be made publicly available.