MALIBA-AI Logo

Bambara ASR Leaderboard

Where Are We at with Automatic Speech Recognition for the Bambara Language ?

MALIBA-AI Dataset Repository GitHub Repo Paper

🏆 Current Best Model: MALIBA-AI/bambara-asr-v3

  • WER: 45.73%
  • CER: 13.45%
  • Combined Score: 36.05%
  • License: Open Source


Main Leaderboard


📊 Custom Ranking Weights
Adjust the sliders below to rank models based on your preference. For example, set WER to 100% and CER to 0% if you only care about Word Error Rate.
0 100
0 100

Current ranking: 50% WER + 50% CER
Rank Model Name WER (%) CER (%) Combined Score (%) License Timestamp
1 🏆 MALIBA-AI/bambara-asr-v3 45.73 13.45 29.59 Open Source 2025-12-23 13:59:12
2 🥈 djelia/asr-v2 47.50 13.56 30.53 Proprietary 2025-12-13 21:20:03
3 🥉 djelia/asr-v1 48.56 13.00 30.78 Proprietary 2025-12-13 21:19:09
4 RobotsMali/soloba-ctc-0.6b-v3 46.76 16.02 31.39 Open Source 2025-12-15 11:56:46
5 RobotsMali/soloni-114m-tdt-ctc-v3-ctc-dec 48.32 14.81 31.57 Open Source 2025-12-15 11:50:57
6 RobotsMali/soloni-114m-tdt-ctc-v3 48.32 14.81 31.57 Open Source 2025-12-13 22:47:24
7 RobotsMali/soloni-114m-tdt-ctc-v2 49.42 15.58 32.50 Open Source 2025-12-13 22:49:43
8 RobotsMali/soloba-ctc-0.6b-v2 48.06 17.19 32.63 Open Source 2025-12-15 11:56:10
9 RobotsMali/soloba-ctc-0.6b-v0.5 49.93 15.33 32.63 Open Source 2025-12-15 11:57:18
10 RobotsMali/soloba-ctc-0.6b-v1.5 52.56 19.93 36.24 Open Source 2025-12-15 11:57:50
11 facebook/mms-1b-all 61.06 14.71 37.88 Open Source 2025-12-13 21:21:07
12 meta/omniASR_LLM_7B 62.57 15.08 38.83 Open Source 2025-12-17 19:04:59
13 sudoping01/bambara-asr-v2 60.33 17.46 38.90 Open Source 2025-12-13 22:17:21
14 RobotsMali/soloba-ctc-0.6b-v1 57.59 20.81 39.20 Open Source 2025-12-15 11:55:46
15 RobotsMali/soloni-114m-tdt-ctc-v0 55.79 22.65 39.22 Open Source 2025-12-13 22:54:22
16 MALIBA-AI/bambara-asr-v1 61.74 17.90 39.82 Open Source 2025-12-13 22:22:44
17 MALIBA-AI/bambara-asr-v2 61.74 17.90 39.82 Open Source 2025-12-14 03:25:54
18 meta/omniASR_LLM_300M 63.32 17.32 40.32 Open Source 2025-12-17 19:07:11
19 Panga-Azazia/bambara-asr-v1.1-0 60.39 22.60 41.49 Proprietary 2025-12-15 12:04:02
20 RobotsMali/stt-bm-quartznet15x5-v2 65.66 18.98 42.32 Open Source 2025-12-13 22:43:12
21 djelia/bm-whisper-large-v2-lora-merged 59.17 25.85 42.51 Proprietary 2025-12-19 19:00:03
22 RobotsMali/soloni-114m-tdt-ctc-v1 61.14 27.69 44.41 Open Source 2025-12-13 22:50:28
23 Panga-Azazia/bambara-asr-ngram 69.13 19.80 44.47 Open Source 2025-12-15 12:02:28
24 Panga-Azazia/bambara-asr 70.00 20.39 45.19 Open Source 2025-12-13 22:15:17
25 meta/omniASR_CTC_1B_v2 69.62 21.93 45.77 Open Source 2025-12-17 19:00:16
26 RobotsMali/soloba-ctc-0.6b-v0 62.93 30.48 46.70 Open Source 2025-12-15 11:54:55
27 meta/omniASR_CTC_3B 72.62 21.80 47.21 Open Source 2025-12-17 19:02:53
28 RobotsMali/stt-bm-quartznet15x5-v1 72.98 21.75 47.37 Open Source 2025-12-13 22:44:00
29 meta/omniASR_LLM_1B 78.31 21.29 49.80 Open Source 2025-12-17 19:04:21
30 meta/omniASR_LLM_CTC_300M 76.87 22.87 49.87 Open Source 2025-12-17 19:07:54
31 meta/omniASR_CTC_7B 74.65 25.47 50.06 Open Source 2025-12-17 19:03:31
32 RobotsMali/stt-bm-quartznet15x5-v0 75.82 25.23 50.53 Open Source 2025-12-13 22:44:55
33 nvidia/parakeet-tdt-0.6b-v3 100.06 49.24 74.65 Open Source 2025-12-17 18:55:17
34 sudoping01/maliba-asr-v0 94.86 71.72 83.29 Open Source 2025-12-14 03:43:28
35 openai/whisper-large-v2 106.84 60.80 83.82 Open Source 2025-12-13 21:51:48
36 nvidia/canary-1b-v2 111.64 60.55 86.09 Open Source 2025-12-17 18:53:59
37 openai/whisper-tiny 112.72 66.61 89.67 Open Source 2025-12-14 13:25:23
38 openai/whisper-small 109.97 75.84 92.91 Open Source 2025-12-14 13:26:31
39 openai/whisper-large-v3 121.06 75.10 98.08 Open Source 2025-12-17 18:57:03
40 openai/whisper-medium 123.18 99.95 111.57 Open Source 2025-12-14 13:25:49
Legend:
🏆 = 1st Place | 🥈 = 2nd Place | 🥉 = 3rd Place

Open Source = Model available on HuggingFace (click name to visit)
Proprietary = Closed source model

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:

    @misc{diallo2025bambaraasr,
      title         = {Where Are We at with Automatic Speech Recognition for the Bambara Language?},
      author        = {Seydou Diallo and Yacouba Diarra and Mamadou K. Keita and Panga Azazia Kamate and Adam Bouno Kampo and Aboubacar Ouattara},
      year          = {2025},
      eprint        = {2602.09785},
      archivePrefix = {arXiv},
      primaryClass  = {cs.CL},
      url           = {https://arxiv.org/abs/2602.09785}
    }


A collaboration between MALIBA-AI, RobotsMali AI4D-LAB, and Djelia

Advancing Speech Recognition Technology for African Languages