Text-to-Speech on SeedTTS
1.43WERBaseline (oracle)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Baseline (oracle)Conditioning source=oracle, Zero-shot protocol=true2026.06 | 1.43 | 68 | 3.66 | |
| Baseline (ASR)Conditioning source=ASR, Zero-shot protocol=true2026.06 | 1.45 | 68 | 3.66 | |
| RTFree Stage 2Training Stage=Stage 2, Zero-shot protocol=true2026.06 | 1.56 | 63 | 3.94 | |
| Reference (GT)2026.07 | 1.82 | 78.9 | — | |
| Base Model2026.07 | 2.22 | 79.4 | — | |
| Unified Guidance2026.07 | 2.45 | 80.6 | — | |
| CosyVoice22026.07 | 2.47 | 75 | — | |
| RTFree Stage 1Training Stage=Stage 1, Zero-shot protocol=true2026.06 | 2.86 | 62 | 3.8 |