Zeroth-Korean
The Zeroth Project releases transcribed Korean recordings together with a pronunciation lexicon and text tools.
Visit official source (opens in a new tab)Research use
Train and evaluate compact Korean recognizers using the published splits and compare text normalization choices.
Limitations
The test set contains 457 utterances from ten speakers. Measure generalization on additional audio for other recording settings.
License and access
- Dataset license
- CC BY 4.0 (opens in a new tab)
- Access status
- Direct access
Download the full archive linked from OpenSLR, which contains speech, transcripts and lexicon files.