Semantic Representation Classification on ARCH Reconstruction Domain
81.25RAVDESS AccuracyOriginal
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Original2025.12 | 81.25 | 99.85 | 47.95 | 59.78 | 62.45 | 39.1 | 65.06 | |
| SpeechTokenizerToken Rate=50, Q=8, Bitrate [kbps]=42025.12 | 78.12 | 99.4 | 45.82 | 57.9 | 54.95 | 36.09 | 62.04 | |
| EncodecToken Rate=75, Q=8, Bitrate [kbps]=62025.12 | 77.78 | 98.41 | 47.62 | 58.74 | 59.45 | 37.98 | 63.33 | |
| DACToken Rate=75, Q=8, Bitrate [kbps]=62025.12 | 77.43 | 99.68 | 47.51 | 58.8 | 62.85 | 40.87 | 64.52 | |
| SACodecToken Rate=75, Q=2, Bitrate [kbps]=1.52025.12 | 75.69 | 99.33 | 45.63 | 59.8 | 58.25 | 39.97 | 63.11 | |
| FACodecToken Rate=80, Q=6, Bitrate [kbps]=4.82025.12 | 70.83 | 98.56 | 47.66 | 59.33 | 56.15 | 36.32 | 61.48 | |
| WavTokenizerToken Rate=75, Q=1, Bitrate [kbps]=0.92025.12 | 68.75 | 99.06 | 45.93 | 59.48 | 58.1 | 33.98 | 60.88 | |
| EncodecToken Rate=75, Q=2, Bitrate [kbps]=1.52025.12 | 66.67 | 95.16 | 46.71 | 58.52 | 57.15 | 36.56 | 60.13 | |
| DACToken Rate=75, Q=2, Bitrate [kbps]=1.52025.12 | 66.67 | 98.3 | 46.41 | 57.42 | 59.55 | 34.18 | 60.42 | |
| SpeechTokenizerToken Rate=50, Q=3, Bitrate [kbps]=1.52025.12 | 57.29 | 97.71 | 43.64 | 54.14 | 51.5 | 36.7 | 56.83 |