Language Identification on VoxLingua33
97.6AccuracyWhisper
Evaluation results
| Method | Links | |
|---|---|---|
| WhisperEvaluation Protocol=MLP2026.02 | 97.6 | |
| Whisper-UEvaluation Protocol=MLP2026.02 | 97.6 | |
| WhisperEvaluation Protocol=kNN2026.02 | 95.8 | |
| Whisper-U-3Evaluation Protocol=MLP2026.02 | 94 | |
| UniWhisperEvaluation Protocol=MLP2026.02 | 89.5 | |
| WavLMEvaluation Protocol=MLP2026.02 | 86.2 | |
| Dasheng 1.2BParameter Count=1.2B2026.01 | 86 | |
| Challenge BaselineDescription=Best among Dasheng-base, data2vec, and Whisper2026.01 | 85.5 | |
| Ensemble (Submission)Ensemble Strategy=Embedding Concatenation2026.01 | 81.7 | |
| HuBERTEvaluation Protocol=MLP2026.02 | 75 | |
| UniWhisperEvaluation Protocol=kNN2026.02 | 71.6 | |
| Whisper-U-3Evaluation Protocol=kNN2026.02 | 62.8 | |
| BEATs (300M) SpeechParameter Count=300M, Pre-training Mixture=Speech-heavy (70:15:15)2026.01 | 61.5 | |
| wav2vec 2.0Evaluation Protocol=MLP2026.02 | 55.8 | |
| Whisper-UEvaluation Protocol=kNN2026.02 | 53.6 | |
| BEATs (300M) BalancedParameter Count=300M, Pre-training Mixture=Balanced (40:30:30)2026.01 | 48 | |
| BEATsEvaluation Protocol=MLP2026.02 | 41.9 | |
| BEATs (90M) iter3Parameter Count=90M, Iteration=32026.01 | 39.8 | |
| WavLMEvaluation Protocol=kNN2026.02 | 29 | |
| CLAP-UEvaluation Protocol=MLP2026.02 | 18.8 | |
| BEATsEvaluation Protocol=kNN2026.02 | 14.8 | |
| CLAP-UEvaluation Protocol=kNN2026.02 | 13.4 | |
| CLAPEvaluation Protocol=MLP2026.02 | 7.7 | |
| HuBERTEvaluation Protocol=kNN2026.02 | 6.3 | |
| CLAPEvaluation Protocol=kNN2026.02 | 3.5 | |
| wav2vec 2.0Evaluation Protocol=kNN2026.02 | 0.6 |