Samromur 21.05

  Authors

Mollberg, David Erik ; et al.show everyone

Mollberg, David Erik ; Jónsson, Ólafur Helgi ; Þorsteinsdóttir, Sunneva ; Guðmundsdóttir, Jóhanna Vigdís ; Steingrímsson, Steinþór ; Magnúsdóttir, Eydís Huld ; Fong, Judy Y ; Borsky, Michal ; Gudnason, Jon

  Item identifier

http://hdl.handle.net/20.500.12537/189

 Project URL

https://github.com/cadia-lvl/samromur

 Date issued

2021-05

 Type

audio, corpus

 Size

145 hours, 8 gb, 100000 utterances

 Language(s)

Icelandic

 Description

This is the first release of the data from the Samrómur collection. The corpus that contains 100,000 (145 hours) validated speech-recordings in Icelandic. The corpus is a result of the crowd-sourcing effort run by the Language and Voice Lab (LVL) at the Reykjavik University, in cooperation with Almannarómur, Center for Language Technology. The recording has started in October 2019 and continues to this day (May 2021). This release has been authorized for release in May 2021. The corpus contains 8,392 different speakers and is split into train, dev, and test subsets with no speaker overlap. Lengths of the sets are: train = 114h, test = 15h, dev = 15h. Each subset contains folders that correspond to speaker IDs, and the audio files inside use the following naming convention: {speaker_ID}-{utterance_ID}.flac. The average recording length is 5.2 seconds. Þetta er fyrsta útgáfan af gögnum úr safni Samróms. Útgáfan inniheldur 100.000 (145 klst.) staðfestar talupptökur á íslensku. Málheildin er afrakstur lýðvistunar (e. crowd sourcing) á vegum Mál- og raddtæknistofu (LVL) við Gervigreindarsetur Háskólans í Reykjavík, í samvinnu við Almannaróm, miðstöð máltækni á Íslandi. Upptökur hófust í október 2019 og standa enn yfir (mai 2021). Hópurinn inniheldur 8.392 mismunandi raddir og hefur verið skipt í þjálfunar- (train), þróunar- (dev) og prófunarsett (test). Lengd settanna er: þjálfunarsett = 114 klst, prófunarsett = 15 klst, þróunarsett = 15 klst. Hvert sett inniheldur möppur sem samsvara auðkenni raddar. Hljóðskrárnar nota eftirfarandi nafnareglur: {speaker_ID}-{utterance_ID}.flac. Meðallengd upptöku er 5,2 sekúnda.

 Publisher

Reykjavík University

 Acknowledgement

Ministry of Education, Science and Culture

Project code: Data recording using Eyra/Samrómur (H1)

Project name: Language Technology for Icelandic 2019-2023

 Subject(s)

audio corpus speech recognition automatic speech recognition

 Collection(s)

Clarin IS

Sýna fulla færslu atriðis