Audio Compression on VCTK
2.64PESQBaseline
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BaselineQuantizer=3 Vanilla RVQ, Down-sampling steps=480, Tokens per second=992024.06 | 2.64 | 0.83 | |
| LLM-CodecDown-sampling steps=480, Tokens per second=572024.06 | 2.55 | 0.82 | |
| BaselineQuantizer=2 Multi-scale RVQ, Down-sampling steps=480, Tokens per second=412024.06 | 2.22 | 0.79 | |
| Encodec_24kQuantizer=3 Vanilla RVQ, Down-sampling steps=320, Tokens per second=2252024.06 | 2.18 | 0.79 | |
| BaselineQuantizer=1 VQ, Down-sampling steps=480, Tokens per second=332024.06 | 2.01 | 0.76 | |
| DAC_16kQuantizer=3 Vanilla RVQ, Down-sampling steps=320, Tokens per second=1502024.06 | 1.76 | 0.78 |