Speech LMs : Wolof
Wolof HuBERT, ASR and speech translation
Official site consultedAI: benchmarks and research
Last verified: 1 October 2026
Paper by Yaya Sy, Dioula Doucouré, Christophe Cerisara and Irina Illina presenting a first speech language model for Wolof. A HuBERT model fine-tuned by continued pretraining improves speech recognition, and the encoder is plugged into a Wolof LLM for speech translation.
Services
- Wolof HuBERT (speech recognition)
- Speech translation
Access requirements
The authors state that models and code will be openly shared, with no link on the page.
Sources
- Speech LMs : Wolof (new tab), verified on 1 October 2026Official site consulted