Fact · Transcription
A 2025 benchmark of speech models for African languages lists the NCHLT corpus as a source of read speech in Zulu, Xhosa and Afrikaans.
Checked against its source on October 17, 2026
In the source's own words
“Common Voice Swahili, Luganda, Kinyarwanda Read NCHLT corpus Zulu, Xhosa, Afrikaans Read”
Benchmarking Automatic Speech Recognition Models for African Languages, arXiv, 2025, Table 1, details of the datasets and languages.
Open the original- What it measures
- Table 1 of the benchmark lists each dataset with its languages and speech style. Zulu and Xhosa appear only through this corpus in the table, and the same paper lists Zulu and Xhosa text corpora for language-model training.