A multilingual evaluation benchmark for automatic speech recognition
covering four under-served languages of the Horn of Africa:
Amharic, Oromo, Somali, and Tigrinya. Each language
ships 1,000 evaluation utterances drawn from spontaneous
interview-style speech with reference transcripts post-edited and
QC-validated by native-speaker annotators.
Total: 4000 utterances, 15.44 hours, 975 distinct speakers (interview proxy).