Audio Reconstruction on LibriSpeech (test-clean test-other)
0.9888CVUTokenBridge
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TokenBridgeCodebook Size=2562026.02 | 0.9888 | 1.1802 | 0.2881 | |
| TokenBridgeCodebook Size=10242026.02 | 0.9888 | 1.1801 | 0.3124 | |
| TokenBridgeCodebook Size=40962026.02 | 0.9887 | 1.2249 | 0.316 | |
| TokenBridgeCodebook Size=163842026.02 | 0.9887 | 1.2328 | 0.3008 | |
| FSPCodebook Size=2562026.02 | 0.1372 | 2.2653 | 0.9083 | |
| VP-VAECodebook Size=2562026.02 | 0.1352 | 2.2718 | 0.91 | |
| FSQCodebook Size=2562026.02 | 0.096 | 1.2763 | 0.7973 | |
| FSPCodebook Size=10242026.02 | 0.081 | 2.4458 | 0.9191 | |
| VP-VAECodebook Size=10242026.02 | 0.0687 | 2.3826 | 0.916 | |
| VP-VAECodebook Size=40962026.02 | 0.0564 | 2.4499 | 0.9259 | |
| FSQCodebook Size=10242026.02 | 0.0513 | 2.1738 | 0.9001 | |
| VP-VAECodebook Size=163842026.02 | 0.0455 | 2.5758 | 0.9339 | |
| FSPCodebook Size=40962026.02 | 0.0368 | 2.4712 | 0.9273 | |
| FSQCodebook Size=40962026.02 | 0.0294 | 2.3821 | 0.9182 | |
| FSPCodebook Size=163842026.02 | 0.0205 | 2.5699 | 0.9324 | |
| FSQCodebook Size=163842026.02 | 0.0149 | 2.4356 | 0.9225 | |
| SimVQCodebook Size=2562026.02 | 0.004 | 1.2524 | 0.4112 | |
| VQ-VAECodebook Size=10242026.02 | 0.001 | 1.4005 | 0.3628 | |
| SimVQCodebook Size=10242026.02 | 0.001 | 1.3213 | 0.4025 | |
| SimVQCodebook Size=40962026.02 | 0.0002 | 1.2519 | 0.3972 | |
| SimVQCodebook Size=163842026.02 | 0.0001 | 1.2477 | 0.4436 |