Bambara ASR Leaderboard
Evaluating Automatic Speech Recognition for Bambara Language
A collaboration between MALIBA-AI, RobotsMali AI4D-LAB, and Djelia
🏆 Current Best Model: MALIBA-AI/bambara-whisper-base
- WER: 22.64%
- CER: 10.94%
- Combined Score: 16.79%
Main Leaderboard
| Model Name | WER (%) | CER (%) | Combined Score (%) | Timestamp |
|---|---|---|---|---|
| 🏆 MALIBA-AI/bambara-whisper-base | 22.64 | 10.94 | 16.79 | 2025-03-15 10:30:45 |
| 🥈 OpenAI/whisper-large-v3 | 32.64 | 15.94 | 24.29 | 2025-03-15 10:30:45 |
| 🥉 Meta/seamless-m4t-v2 | 41.56 | 21.34 | 31.49 | 2025-03-15 10:30:45 |
| djelia/asr-v2 | 56.13 | 18.81 | 44.94 | 2025-11-25 03:30:47 |
| sed/test/v2 | 62.17 | 18.60 | 49.10 | 2025-11-25 03:03:01 |
Citation
If you use the Bambara ASR benchmark for your scientific publication, or if you find the resources in this leaderboard useful, please cite our work:
@article{bambara_asr_benchmark_2025,
title={Bambara ASR Benchmark: Evaluating Speech Recognition for a Low-Resource African Language},
author={MALIBA-AI Team and RobotsMali AI4D-LAB and Djelia},
journal={arXiv preprint},
year={2025},
url={https://huggingface.co/datasets/MALIBA-AI/bambara-speech-recognition-leaderboard}
}
About the Collaboration
This benchmark is a collaborative effort between:
- MALIBA-AI - Mission: "No Malian Language Left Behind" - Empowering Mali's linguistic diversity through AI innovation
- RobotsMali AI4D-LAB - Advancing AI research and development in Mali
- Djelia - Preserving and promoting African languages through technology
A collaboration between MALIBA-AI, RobotsMali AI4D-LAB, and Djelia
Advancing Speech Recognition Technology for African Languages