Speech Understanding on DASB
47.4Error Rate (ER)EnCodec
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| EnCodecFPS/TPS=50/150, CN/CS=3/1024, BR (kbps)=1.52026.02 | 47.4 | 15.5 | 79.3 | |
| DACFPS/TPS=50/150, CN/CS=3/1024, BR (kbps)=1.52026.02 | 48.9 | 17.8 | 68.8 | |
| WavTokenizerFPS/TPS=40/40, CN/CS=1/4096, BR (kbps)=0.482026.02 | 51.1 | 19.4 | 65.3 | |
| MimiFPS/TPS=12.5/100, CN/CS=8/2048, BR (kbps)=1.12026.02 | 54.3 | 19.7 | 92.2 | |
| SiTok (CN = 1)FPS/TPS=12.5/12.5, CN/CS=1/65536, BR (kbps)=0.22026.02 | 63.5 | 13.8 | 96.9 | |
| SiTok (CN = 4)FPS/TPS=12.5/50, CN/CS=4/16384, BR (kbps)=0.72026.02 | 64.4 | 8.59 | 97.7 |