FSDKaggle2018
Models assign a sound event label to a Freesound recording.
The original challenge ranks up to three proposed labels for each clip.
- Made by
- Eduardo Fonseca and colleagues
- Released
- March 2018
- Kind
- Audio understanding and voice
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Prediction
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Audio and speech
Cochlear.ai_1 tops the leaderboard with 0.95
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.
- 0.95 Cochlear.ai_1
- 0.95 CNN - Iterative Self-Verification
- 0.95 Stacked CNN-CRNN (8 Models)
- 0.95 Cochlear.ai_2
- 0.95 DCASE2018 Meta-learning system
- 0.95 NTU_ensemble8
- 0.95 Stacked CNN-CRNN (4 Models)
- 0.95 NTU_en8_augment_test
- 0.94 CNN on Raw-Audio and Spectrogram
- 0.94 Kuaiyu tagging system
- 0.94 Ensemble of convolutional neural networks for general purpose audio tagging
- 0.94 CNN Ensemble based on Multiple Features
- 0.94 Cochlear.ai_3
- 0.94 Shan DBSonics approach
- 0.93 3 CNN L1 Stacked Fused Spectral XGBoost L2
- 0.93 2 CNN results geometrically averaged
- 0.93 Fusion of 4 CNN
- 0.93 NTU_labelsmoothing
- 0.92 NTU_bgnormalization
- 0.92 2 average pooling
- 0.92 2 max pooling
- 0.92 ConResNet
- 0.91 Multi-level attention model on fine-tuned AudioSet features
- 0.91 VGG Style CNN with 3 channel input
- 0.90 Surrey baseline CNN 8 layers
- 0.88 BCNN_WaveNet
- 0.87 2ModEnsem
- 0.86 Surrey baseline CNN 4 layers
- 0.79 CRNN
- 0.70 DCASE2018 Task2 CRNN RM3
- 1 Cochlear.ai_1 Jeong_COCAI_task2_1 0.95
- 2 CNN - Iterative Self-Verification Dorfer_CPJKU_task2_1 0.95
- 3 Stacked CNN-CRNN (8 Models) Iqbal_Surrey_task2_2 0.95
- 4 Cochlear.ai_2 Jeong_COCAI_task2_2 0.95
- 5 DCASE2018 Meta-learning system Kele_NUDT_task2_1 0.95
- 6 NTU_ensemble8 Nguyen_NTU_task2_1 0.95
- 7 Stacked CNN-CRNN (4 Models) Iqbal_Surrey_task2_1 0.95
- 8 NTU_en8_augment_test Nguyen_NTU_task2_4 0.95
- 9 CNN on Raw-Audio and Spectrogram Wilhelm_UKON_task2_1 0.94
- 10 Kuaiyu tagging system Wei_Kuaiyu_task2_2 0.94
- 11 Ensemble of convolutional neural networks for general purpose audio tagging Pantic_ETF_task2_1 0.94
- 12 CNN Ensemble based on Multiple Features Wilkinghoff_FKIE_task2_1 0.94
- 13 Cochlear.ai_3 Jeong_COCAI_task2_3 0.94
- 14 Shan DBSonics approach Shan_DBSonics_task2_1 0.94
- 15 3 CNN L1 Stacked Fused Spectral XGBoost L2 Chakraborty_IBM_Task2_1 0.93
- 16 2 CNN results geometrically averaged Chakraborty_IBM_Task2_2 0.93
- 17 Fusion of 4 CNN Agafonov_ITMO_task2_2 0.93
- 18 NTU_labelsmoothing Nguyen_NTU_task2_2 0.93
- 19 NTU_bgnormalization Nguyen_NTU_task2_3 0.92
- 20 2 average pooling Khadkevich_FB_task2_1 0.92
- 21 2 max pooling Khadkevich_FB_task2_2 0.92
- 22 ConResNet Kim_GIST_WisenetAI_task2_4 0.92
- 23 Multi-level attention model on fine-tuned AudioSet features Xu_Aalto_task2_2 0.91
- 24 VGG Style CNN with 3 channel input Chakraborty_IBM_Task2_judges_award 0.91
- 25 Surrey baseline CNN 8 layers Baseline_Surrey_task2_1 0.90
- 26 BCNN_WaveNet Zhesong_PKU_task2_1 0.88
- 27 2ModEnsem Han_NPU_task2_1 0.87
- 28 Surrey baseline CNN 4 layers Baseline_Surrey_task2_2 0.86
- 29 CRNN Hanyu_BUPT_task2 0.79
- 30 DCASE2018 Task2 CRNN RM3 Colangelo_RM3_task2_1 0.70
Source: Official FSDKaggle2018 leaderboard
| # | Entry | Score | Setting | Submitted by | Date | Source |
|---|---|---|---|---|---|---|
| 1 | Cochlear.ai_1 | 0.95 | Jeong_COCAI_task2_1 | Jeong_COCAI | Official leaderboard | |
| 2 | CNN - Iterative Self-Verification | 0.95 | Dorfer_CPJKU_task2_1 | Dorfer_CPJKU | Official leaderboard | |
| 3 | Stacked CNN-CRNN (8 Models) | 0.95 | Iqbal_Surrey_task2_2 | Iqbal_Surrey | Official leaderboard | |
| 4 | Cochlear.ai_2 | 0.95 | Jeong_COCAI_task2_2 | Jeong_COCAI | Official leaderboard | |
| 5 | DCASE2018 Meta-learning system | 0.95 | Kele_NUDT_task2_1 | Kele_NUDT | Official leaderboard | |
| 6 | NTU_ensemble8 | 0.95 | Nguyen_NTU_task2_1 | Nguyen_NTU | Official leaderboard | |
| 7 | Stacked CNN-CRNN (4 Models) | 0.95 | Iqbal_Surrey_task2_1 | Iqbal_Surrey | Official leaderboard | |
| 8 | NTU_en8_augment_test | 0.95 | Nguyen_NTU_task2_4 | Nguyen_NTU | Official leaderboard | |
| 9 | CNN on Raw-Audio and Spectrogram | 0.94 | Wilhelm_UKON_task2_1 | Wilhelm_UKON | Official leaderboard | |
| 10 | Kuaiyu tagging system | 0.94 | Wei_Kuaiyu_task2_2 | Wei_Kuaiyu | Official leaderboard | |
| 11 | Ensemble of convolutional neural networks for general purpose audio tagging | 0.94 | Pantic_ETF_task2_1 | Pantic_ETF | Official leaderboard | |
| 12 | CNN Ensemble based on Multiple Features | 0.94 | Wilkinghoff_FKIE_task2_1 | Wilkinghoff_FKIE | Official leaderboard | |
| 13 | Cochlear.ai_3 | 0.94 | Jeong_COCAI_task2_3 | Jeong_COCAI | Official leaderboard | |
| 14 | Shan DBSonics approach | 0.94 | Shan_DBSonics_task2_1 | Shan_DBSonics | Official leaderboard | |
| 15 | 3 CNN L1 Stacked Fused Spectral XGBoost L2 | 0.93 | Chakraborty_IBM_Task2_1 | Chakraborty_IBM | Official leaderboard | |
| 16 | 2 CNN results geometrically averaged | 0.93 | Chakraborty_IBM_Task2_2 | Chakraborty_IBM | Official leaderboard | |
| 17 | Fusion of 4 CNN | 0.93 | Agafonov_ITMO_task2_2 | Agafonov_ITMO | Official leaderboard | |
| 18 | NTU_labelsmoothing | 0.93 | Nguyen_NTU_task2_2 | Nguyen_NTU | Official leaderboard | |
| 19 | NTU_bgnormalization | 0.92 | Nguyen_NTU_task2_3 | Nguyen_NTU | Official leaderboard | |
| 20 | 2 average pooling | 0.92 | Khadkevich_FB_task2_1 | Khadkevich_FB | Official leaderboard | |
| 21 | 2 max pooling | 0.92 | Khadkevich_FB_task2_2 | Khadkevich_FB | Official leaderboard | |
| 22 | ConResNet | 0.92 | Kim_GIST_WisenetAI_task2_4 | Kim_GIST_WisenetAI | Official leaderboard | |
| 23 | Multi-level attention model on fine-tuned AudioSet features | 0.91 | Xu_Aalto_task2_2 | Xu_Aalto | Official leaderboard | |
| 24 | VGG Style CNN with 3 channel input | 0.91 | Chakraborty_IBM_Task2_judges_award | Chakraborty_IBM | Official leaderboard | |
| 25 | Surrey baseline CNN 8 layers | 0.90 | Baseline_Surrey_task2_1 | Baseline_Surrey | Official leaderboard | |
| 26 | BCNN_WaveNet | 0.88 | Zhesong_PKU_task2_1 | Zhesong_PKU | Official leaderboard | |
| 27 | 2ModEnsem | 0.87 | Han_NPU_task2_1 | Han_NPU | Official leaderboard | |
| 28 | Surrey baseline CNN 4 layers | 0.86 | Baseline_Surrey_task2_2 | Baseline_Surrey | Official leaderboard | |
| 29 | CRNN | 0.79 | Hanyu_BUPT_task2 | Hanyu_BUPT | Official leaderboard | |
| 30 | DCASE2018 Task2 CRNN RM3 | 0.70 | Colangelo_RM3_task2_1 | Colangelo_RM3 | 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 (30 entries).
- Pass Models scored 30 entries on its leaderboard, none a model we track.
- Note Newest score No scores yet.
- Note Frontier models No models scored yet.
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on dcase.community: 30 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.