Uzbek Speech Corpus
Models transcribe held-out recordings of people reading Uzbek. Scorers compare the transcripts with the reference words.
NavAI reports word error rates for 34 systems in its USC column. Asliddin’s separate comparison tests Whisper and Wav2Vec2.
- Made by
- ISSAI and Tashkent University of Information Technologies
- Released
- July 2021
- Kind
- Speech recognition and translation
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Language
- Uzbek
- Interaction
- Single turn
- Items from
- Written by crowdworkers
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Audio and speech
ai-sage/GigaAM-Multilingual (Large) tops the leaderboard with 8.74%
Best score per entry from NavAI’s leaderboard, read on November 20th, 2026. 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 ai-sage/GigaAM-Multilingual (Large) Single-pass decoding 8.74%
- 2 navai-uz/whisper-medium-uzbek 9.70%
- 3 ai-sage/GigaAM-Multilingual (220M) Single-pass decoding 11.4%
- 4 islomov/rubaistt_v2_medium FLEURS test data in training, USC train-only 11.6%
- 5 navai-uz/whisper-small-uzbek 11.8%
- 6 nvidia/stt_uz_fastconformer_hybrid_large_pc Single-pass decoding 12.1%
- 7 uzinfocom-edu-ai/asr-uz-fastconformer-large Single-pass decoding 12.9%
- 8 Kotib/uzbek_stt_v1 Training data not disclosed 13.3%
- 9 navai-uz/whisper-base-uzbek 14.9%
- 10 mrmuminov/whisper-small-uz 17.3%
- 11 navai-uz/whisper-tiny-uzbek 18.7%
- 12 facebook/seamless-m4t-v2-large Single-pass decoding 19.2%
- 13 OvozifyLabs/whisper-small-uz-v1 19.4%
- 14 Abduqayum/whisper-uzbek-medium-callcenter 20.6%
- 15 aisha-org/Whisper-Uzbek 21.2%
- 16 Yorkinjon/whisper-medium-uzbek-ynv1 22.4%
- 17 abduaziz/whisper-small-uzbek 27%
- 18 Yorkinjon/whisper-small-uzbek-ynv3 27%
- 19 Yorkinjon/whisper-small-uzbek-ynv2 28%
- 20 blackhole33/whisper-small-uz 28.4%
- 21 oyqiz/uzbek_stt Single-pass decoding 29.9%
- 22 Yorkinjon/whisper-small-uzbek-yn 30.1%
- 23 mustafoyev202/whisper-uz 30.8%
- 24 tukhtashevshohruh/whisper-small-uz 34.1%
- 25 firdavsus/whisper-small-uz-ru 34.2%
- 26 jmshd/whisper-uz 36.7%
- 27 blackhole33/whisper-small-uzb-commonvoice 38.3%
- 28 azamat45/whisper-small-uz 39%
- 29 mirodil/whisper-small-uzbek 40.1%
- 30 mirodil/whisper-tiny-uzbek 41.4%
- 31 Inomjonov/mironshoh-whisper-medium-uz-10k 41.7%
- 32 mrmuminov/whisper-tiny-uz 44%
- 33 Xojakbar/whisper-small-uz 55.6%
- 34 Saidakbar01/whisper-uz-finetuned 75.3%
Source: NavAI’s leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | ai-sage/GigaAM-Multilingual (Large) | 8.74% | Single-pass decoding, uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 2 | navai-uz/whisper-medium-uzbek | 9.70% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 3 | ai-sage/GigaAM-Multilingual (220M) | 11.4% | Single-pass decoding, uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 4 | islomov/rubaistt_v2_medium | 11.6% | FLEURS test data in training, USC train-only, uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 5 | navai-uz/whisper-small-uzbek | 11.8% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 6 | nvidia/stt_uz_fastconformer_hybrid_large_pc | 12.1% | Single-pass decoding, uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 7 | uzinfocom-edu-ai/asr-uz-fastconformer-large | 12.9% | Single-pass decoding, uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 8 | Kotib/uzbek_stt_v1 | 13.3% | Training data not disclosed, uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 9 | navai-uz/whisper-base-uzbek | 14.9% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 10 | mrmuminov/whisper-small-uz | 17.3% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 11 | navai-uz/whisper-tiny-uzbek | 18.7% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 12 | facebook/seamless-m4t-v2-large | 19.2% | Single-pass decoding, uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 13 | OvozifyLabs/whisper-small-uz-v1 | 19.4% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 14 | Abduqayum/whisper-uzbek-medium-callcenter | 20.6% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 15 | aisha-org/Whisper-Uzbek | 21.2% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 16 | Yorkinjon/whisper-medium-uzbek-ynv1 | 22.4% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 17 | abduaziz/whisper-small-uzbek | 27% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 18 | Yorkinjon/whisper-small-uzbek-ynv3 | 27% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 19 | Yorkinjon/whisper-small-uzbek-ynv2 | 28% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 20 | blackhole33/whisper-small-uz | 28.4% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 21 | oyqiz/uzbek_stt | 29.9% | Single-pass decoding, uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 22 | Yorkinjon/whisper-small-uzbek-yn | 30.1% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 23 | mustafoyev202/whisper-uz | 30.8% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 24 | tukhtashevshohruh/whisper-small-uz | 34.1% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 25 | firdavsus/whisper-small-uz-ru | 34.2% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 26 | jmshd/whisper-uz | 36.7% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 27 | blackhole33/whisper-small-uzb-commonvoice | 38.3% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 28 | azamat45/whisper-small-uz | 39% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 29 | mirodil/whisper-small-uzbek | 40.1% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 30 | mirodil/whisper-tiny-uzbek | 41.4% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 31 | Inomjonov/mironshoh-whisper-medium-uz-10k | 41.7% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 32 | mrmuminov/whisper-tiny-uz | 44% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 33 | Xojakbar/whisper-small-uz | 55.6% | uzbek_text_norm v0.3.0 (greedy) | Official leaderboard | |
| 34 | Saidakbar01/whisper-uz-finetuned | 75.3% | uzbek_text_norm v0.3.0 (greedy) | 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 All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the official leaderboard (34 entries).
- Pass Models scored 34 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 NavAI: 34 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.