Technology
Initiatives to Train Artificial Intelligence in 8 Indigenous Ethiopian Languages
BBC · 5 hours ago · Read the Full Story on BBC
AI SUMMARY
Addis Ababa University and private institutions are collecting audio data across various regions and dialects to train artificial intelligence models in eight indigenous Ethiopian languages.
- Addis Ababa University is training artificial intelligence models in eight indigenous languages: Ge'ez, Amharic, Afaan Oromo, Wolayttatto, Tigrinya, Sidaamu Afoo, Afar, and Somali.
- Up to 100 hours of audio have been collected for each of these languages through field recordings that ensure gender balance and regional representation.
- Gira Solutions, formerly known as iCog Labs, is conducting a project named 'Layu' to collect extensive speech data for local languages.
- The 'Layu' project has gathered over 550 hours of Amharic speech covering five major dialects from Shewa, Addis Ababa, Gondar, Gojjam, and Wello.
- Gira Solutions is also collecting Afaan Oromo speech data from regions including Harar, Arsi, and Jimma in collaboration with local universities.
- Experts note that training an AI model requires thousands of hours of audio and high-quality data to ensure accurate language processing and pattern recognition.
Related
Ethio Telecom Deploys First AI-Powered Mobile Network Technology
Capital Ethiopia · 1 hour ago
Ethiopia’s FaydaPass Digital Wallet Could Gain Blockchain-Backed Credentials
Biometric Update · 2 days ago
Turning Policy Into Practice Through The UAE’s AI Blueprint
The Reporter · 6 days ago
Communications Authority Regulatory Costs to Be Weighed Against Benefits
BirrMetrics · last week