Voice Conversion on Seed-TTS zh (test)
1.33WERGround Truth
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Ground TruthInference mode=Offline, Task setting=Zero-shot2026.04 | 1.33 | — | — | — | |
| Ground TruthZero-shot=true, Streaming=true2026.04 | 1.33 | — | — | — | |
| Seed-VCtiny#Param. (Total)=262M, #Param. (Converter)=27M, Inference mode=Offline, Task setting=Zero-shot2026.04 | 1.79 | 0.6 | 3.08 | 0.069 | |
| X-VC#Param. (Total)=539M, #Param. (Converter)=44M, Inference mode=Offline, Task setting=Zero-shot2026.04 | 1.99 | 0.73 | 2.69 | 0.014 | |
| Seed-VCsmall#Param. (Total)=296M, #Param. (Converter)=98M, Inference mode=Offline, Task setting=Zero-shot2026.04 | 2.52 | 0.73 | 2.73 | 0.161 | |
| X-VC#Param. Total=539M, #Param. Converter=44M, Latency (T_model)=240, Latency (T_compute)=58.17, Zero-shot=true, Streaming=true2026.04 | 2.65 | 0.72 | 2.35 | — | |
| Seed-VCtiny#Param. Total=262M, #Param. Converter=27M, Latency (T_model)=380, Latency (T_compute)=120.75, Zero-shot=true, Streaming=true2026.04 | 3.36 | 0.6 | 2.46 | — | |
| MeanVC#Param. (Total)=47M, #Param. (Converter)=14M, Inference mode=Offline, Task setting=Zero-shot2026.04 | 3.89 | 0.73 | 2.81 | 0.094 | |
| MeanVC#Param. Total=47M, #Param. Converter=14M, Latency (T_model)=250, Latency (T_compute)=32.12, Zero-shot=true, Streaming=true2026.04 | 4.89 | 0.72 | 2.22 | — |