Lahjoita puhetta ASR test
Models transcribe spontaneous Finnish speech on the official ten-hour test. Human transcripts supply the reference words.
Getman and colleagues reused the official test in 2024 to compare pretrained wav2vec models. The separate one-hour sample has four transcribers.
- Made by
- Aalto University and University of Helsinki
- Released
- March 2022
- Kind
- Speech recognition and translation
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- Kept private
- Measures
- Capability
- Language
- Finnish
- Interaction
- Single turn
- Items from
- Written by crowdworkers
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Audio and speech
Wav2Vec2+CTC tops the leaderboard with 21.8%
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 Wav2Vec2+CTC 100h training, external language model data 21.8%
- 2 TDNN (large-GMM alignments) 1600h training, external language model data 23.8%
- 3 initial-TDNN 1600h training 24%
- 4 semisup-TDNN 100h training, external language model data 28.2%
- 5 AED 1600h training 34.9%
- 6 large-GMM 1600h training 39%
- 7 initial-GMM 1600h training 40.9%
Source: The Lahjoita puhetta ASR test paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Wav2Vec2+CTC | 21.8% | 100h training, external language model data | Official leaderboard | |
| 2 | TDNN (large-GMM alignments) | 23.8% | 1600h training, external language model data | Official leaderboard | |
| 3 | initial-TDNN | 24% | 1600h training | Official leaderboard | |
| 4 | semisup-TDNN | 28.2% | 100h training, external language model data | Official leaderboard | |
| 5 | AED | 34.9% | 1600h training | Official leaderboard | |
| 6 | large-GMM | 39% | 1600h training | Official leaderboard | |
| 7 | initial-GMM | 40.9% | 1600h training | 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 The tasks are private.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the benchmark’s paper (7 entries).
- Pass Models scored 7 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 arxiv.org: 7 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.