Difference between revisions of "2025:Song Deepfake Detection Results"

From MIREX Wiki
(Baselines)
(Results)
Line 9: Line 9:
 
! style="text-align:right;" | EER on WildSVDD test_B
 
! style="text-align:right;" | EER on WildSVDD test_B
 
! style="text-align:right;" | EER on SONICS test
 
! style="text-align:right;" | EER on SONICS test
|- style="background-color:#FF0;"
 
 
| Baseline1
 
| Baseline1
 
| RawNet - Mixtures
 
| RawNet - Mixtures
Line 24: Line 23:
 
| style="text-align:right;" | 1.75
 
| style="text-align:right;" | 1.75
 
|-
 
|-
 +
|- style="background-color:#FF0;"
 
| Baseline3
 
| Baseline3
 
| Wav2vec2 + AASIST - Mixtures
 
| Wav2vec2 + AASIST - Mixtures

Revision as of 23:27, 10 September 2025

Results

System Methods Used Trained on the training set of EER on WildSVDD test_A EER on WildSVDD test_B EER on SONICS test Baseline1 RawNet - Mixtures WildSVDD 9.44 30.85 24.36
Baseline2 SpecTTTra - Mixtures SONICS 48.92 43.96 1.75
Baseline3 Wav2vec2 + AASIST - Mixtures WildSVDD + SONICS 6.14 20.82 2.05