KSAA 2026 Automatic Diacritization of Speech Dictation
Models restore Arabic vowel marks from a recording and its transcript. Organizers rank systems by word error rate with case endings and unmarked characters included.
Thaka reported 23.26% word error with CATT-Whisper on the blind test. The organizers ranked five teams and reported three baselines.
- Released
- December 2025
- Kind
- Speech recognition and translation
- Level
- School
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- Public, plus a private test set
- Measures
- Capability
- Subject
- Languages and literature
- Language
- Arabic
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- A host
- In views
- Audio and speech
Thaka tops the leaderboard with 23.3%
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.
- 1 Thaka 23.3%
- 2 TantaArabNLP 24.4%
- 3 Abjad AI 25.3%
- 4 Eraserhead 30.4%
- 5 Fine-tune text + ASR Organizer baseline 31.8%
- 6 Fine-Tashkeel 34.5%
- 7 text + ASR Organizer baseline 40.2%
- 8 text only Organizer baseline 49.9%
Source: The KSAA 2026 Automatic Diacritization of Speech Dictation paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Thaka | 23.3% | Official leaderboard | ||
| 2 | TantaArabNLP | 24.4% | Official leaderboard | ||
| 3 | Abjad AI | 25.3% | Official leaderboard | ||
| 4 | Eraserhead | 30.4% | Official leaderboard | ||
| 5 | Fine-tune text + ASR | 31.8% | Organizer baseline | Official leaderboard | |
| 6 | Fine-Tashkeel | 34.5% | Official leaderboard | ||
| 7 | text + ASR | 40.2% | Organizer baseline | Official leaderboard | |
| 8 | text only | 49.9% | Organizer baseline | Official leaderboard |
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Audio and video
- Video-MMMU In use 300 expert-level lecture videos with questions testing perception, comprehension and adaptation of knowledge. Accuracy.
Qwen 3.8 Max 88.7%
- LVBench In use Hours-long video understanding with multiple-choice questions on reasoning, grounding and summarization.
Gemini 4 Argon 91.7%
- MMVU In use 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy.
Qwen 3.8 Max 82.4%
- MLVU In use Multi-task long video understanding (minutes to hours) with multiple-choice and generation tasks. Accuracy.
Qwen 3.8 Max 90.8%
- FLEURS In use Models transcribe read speech in 102 languages built from FLoRes-101 sentences. Model cards usually report word error rate, so lower scores are better.
Phi-4-multimodal 4.00%
- VoiceBench In use Spoken instructions with speaker, environment and content variation. Voice assistant responses scored by GPT judge or rules.
Qwen3.5-Omni-Plus 93.1%
Data audit
All 8 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access Public tasks, plus a private test set.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the benchmark’s paper (8 entries).
- Pass Models scored 8 entries on its leaderboard, none a model we track.
- Note Newest score No scores yet.
- Note Frontier models No models scored yet.
- Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
- Pass Importer We read the board on codabench.org: 8 entries at the last read, none a model we track.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.