Benchmark Almanac

Amharic ASR Benchmark

boazsew/Ethio-ASR-w2v-bert-2.0-combined-am-uf 5.76% tops its leaderboard Whisper large-v3 151.1% leads the models we track
Not in model cards

Speech models transcribe 1,548 recordings of people reading Amharic prompts. The organizers report character and word error rates.

Dataset.ET lists 13 models in its clean comparison, one trained on the source corpus and two with undisclosed training data.

Made by
Dataset.ET
Released
September 2026
Kind
Speech recognition and translation
Level
Everyday
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
Amharic
In and out
Audio → Text
Interaction
Single turn
Items from
Another benchmark
Refresh
One fixed set
Run by
Anyone
In views
Audio and speech
Last model added
November 2023
Scores refreshed
November 20th, 2026
16
Entries on its own leaderboard, 1 model we track

boazsew/Ethio-ASR-w2v-bert-2.0-combined-am-uf tops the leaderboard with 5.76% Whisper large-v3 leads with 151.1%

Best score per entry from the benchmark's own 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. The best model we track is Whisper large-v3, at 151.1%. This view shows only models we track, each at its best.

  1. 1 boazsew/Ethio-ASR-w2v-bert-2.0-combined-am-uf Trained on this dataset 5.76%
  2. 2 boazsew/Ethio-ASR-w2v-bert-2.0-uf Clean 9.24%
  3. 3 badrex/Ethio-ASR-amharic Clean 9.46%
  4. 4 badrex/Ethio-ASR-multilingual-600M Clean 9.91%
  5. 5 b1n1yam/shook-medium-amharic-2k Clean 11.5%
  6. 6 aadel4/omniASR-CTC-1B-v2 Clean 12.3%
  7. 7 badrex/Ethio-ASR-multilingual-1B Clean 13.1%
  8. 8 badrex/Ethio-ASR-multilingual-94M Clean 15.4%
  9. 9 badrex/Ethio-ASR-multilingual-300M Clean 15.9%
  10. 10 boazsew/Ethio-ASR-afrihubert-uf Clean 16.5%
  11. 11 b1n1yam/qwen3-asr-0.6b-amharic-mixed Undisclosed training data 18.4%
  12. 12 b1n1yam/qwen3-asr-0.6b-amharic-gold-silver Undisclosed training data 18.6%
  13. 13 facebook/seamless-m4t-v2-large Clean 23.1%
  14. 14 agkphysics/wav2vec2-large-xlsr-53-amharic Clean 25%
  15. 15 b1n1yam/shook-tiny-amharic-stage2-polish Clean 27.4%
  16. 16 Whisper large-v3 Clean 151.1%

Source: Official Amharic ASR Benchmark leaderboard

Source: Official Amharic ASR Benchmark leaderboard

#EntryScore ±SettingDateLinkSource
1 boazsew/Ethio-ASR-w2v-bert-2.0-combined-am-uf 5.76% ±0.4 Trained on this dataset huggingface.co Official leaderboard
2 boazsew/Ethio-ASR-w2v-bert-2.0-uf 9.24% ±0.5 Clean huggingface.co Official leaderboard
3 badrex/Ethio-ASR-amharic 9.46% 9 to 10 Clean huggingface.co Official leaderboard
4 badrex/Ethio-ASR-multilingual-600M 9.91% ±0.5 Clean huggingface.co Official leaderboard
5 b1n1yam/shook-medium-amharic-2k 11.5% 10.9 to 12 Clean huggingface.co Official leaderboard
6 aadel4/omniASR-CTC-1B-v2 12.3% ±0.5 Clean huggingface.co Official leaderboard
7 badrex/Ethio-ASR-multilingual-1B 13.1% ±0.6 Clean huggingface.co Official leaderboard
8 badrex/Ethio-ASR-multilingual-94M 15.4% 14.8 to 16 Clean huggingface.co Official leaderboard
9 badrex/Ethio-ASR-multilingual-300M 15.9% 15.3 to 16.5 Clean huggingface.co Official leaderboard
10 boazsew/Ethio-ASR-afrihubert-uf 16.5% 15.9 to 17.1 Clean huggingface.co Official leaderboard
11 b1n1yam/qwen3-asr-0.6b-amharic-mixed 18.4% 17.6 to 19.1 Undisclosed training data huggingface.co Official leaderboard
12 b1n1yam/qwen3-asr-0.6b-amharic-gold-silver 18.6% ±0.8 Undisclosed training data huggingface.co Official leaderboard
13 facebook/seamless-m4t-v2-large 23.1% 22 to 24.2 Clean huggingface.co Official leaderboard
14 agkphysics/wav2vec2-large-xlsr-53-amharic 25% ±0.5 Clean huggingface.co Official leaderboard
15 b1n1yam/shook-tiny-amharic-stage2-polish 27.4% ±0.9 Clean huggingface.co Official leaderboard
16 Whisper large-v3 151.1% 148.8 to 153.4 Clean Nov 2023 huggingface.co Official leaderboard
#ModelScore ±SettingDateLinkSource
1 Whisper large-v3 151.1% 148.8 to 153.4 Clean Nov 2023 huggingface.co Official leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
0% 20% 40% 60% 80% 100% 2024 Whisper large-v3 0% 20% 40% 60% 80% 100% 2024 Whisper large-v3

Source: Official Amharic ASR Benchmark 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

All 695 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

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 3 years ago, in November 2023, for Whisper large-v3. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (16 entries).
  • Pass Models scored 1 model we track, plus 15 other entries on its leaderboard.
  • Warning Newest score The newest came 3 years ago, in November 2023, for Whisper large-v3.
  • Note Frontier models None of its models are on the capabilities index.
  • Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
  • Pass Importer We read the board on huggingface.co: 1 row for models we track and 15 other entries at the last read.
  • 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.

How we check