This release of data from the Samrómur collection contains focuses on users where Icelandic is not their native tongue. The data is mostly UNVERIFIED. It contains 143,031 (151.8 hours) speech-recordings in Icelandic, of which 4,957 have been verified. 42,000 utterances have been scored with marosijo which indicates if it is likely to be valid or not.

The corpus is a result of the crowd-sourcing effort run by the Language and Voice Lab (LVL) at Reykjavik University, in cooperation with Almannarómur, the Icelandic Center for Language Technology. The recording process has started in October 2019 and ended September 2022. The present edition of the corpus has been authorized for release in September 2022. The aim is to create an open-source speech corpus to enable research and development for Icelandic Language Technology. The corpus consists of audio recordings and a metadata file containing the prompts read by the participants.

Participants are from 6 and up to 80+ years. The distributed audio files are encoded at 16 kHz sampling rate, 16 bit linear PCM, 1 channel, *.flac format. The corpus is split into train, dev and test subsets. Lengths of the sets are: train = 91.4h, test = 29.3h, dev = 31.1h. For more such subsets please use ones from the other Samrómur releases. All demographics are self reported. The dataset contains folders that correspond to speaker IDs, and the audio files inside use the following naming convention: {speaker_ID}-{utterance_ID}.flac.

You can cite the data using the following BibTeX entry:

    @inproceedings{hedstrom2022,
        title={{Samr{\'o}mur L2 22.09}},
        author={Staffan Hedstr{\"o}m, Ragnhei{\dh}ur {\TH}{\'o}rhallsd{\'o}ttir,
            David Erik Mollberg, Thomas Mestrou,  Sm{\'a}ri Freyr Gu{\dh}mundsson,
            {\'O}lafur Helgi J{\'o}nsson, Sunneva {\TH}orsteinsd{\'o}ttir,
            Judy Y. Fong, Eyd{\'\i}s Huld Magn{\'u}sd{\'o}ttir, Jon Gudnason},
        publisher={Reykjavik University: Language and Voice Lab},
        year={2022},
      }