Audio Reconstruction on VGGSound (test)
35.45SI-SDRVanilla
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| VanillaType=AV, Resolution=16 × 128 × 128, #Tokens=11522026.06 | 35.45 | 10.26 | 2.114 | |
| SpectralCodecType=AO, Resolution=80 × 384 (M), #Tokens=3842026.06 | 29.3 | 5.56 | 1.514 | |
| WavTokenizerType=AO, Resolution=98304 × 1 (W), #Tokens=1642026.06 | 24.27 | 6.82 | 1.589 | |
| AVTokType=AV, Resolution=80 × 384 (M), #Tokens=11522026.06 | 23.09 | 5.93 | 1.523 | |
| UniCodecType=AO, Resolution=98304 × 1 (W), #Tokens=3082026.06 | 18.25 | 6.73 | 1.508 |