List your company

Text-to-Speech on SeedTTS en (test)

1.521WERIndexTTS 2
-0.1041610.8656721.835532.80533Aug 22, 2025Sep 15, 2025Oct 10, 2025Nov 4, 2025Nov 29, 2025Dec 24, 2025Jan 18, 2026

Evaluation results

MethodLinks
IndexTTS 2mode=Zero-shot2026.011.521——0.86—
FireRedTTS-2mode=Zero-shot2026.011.652——0.782—
IndexTTS 2.5-RLmode=Reinforcement learning-based fine-tuning2026.011.732——0.847—
IndexTTS 2.5mode=Zero-shot2026.011.889——0.855—
Ground Truthmode=Zero-shot2026.011.897——0.82—
CosyVoice 3mode=Zero-shot2026.012.02——0.811—
Ground Truth#Hz=-, Data (#hours)=-, PT?=false2025.082.1433.523——0.73
MaskGCT#Hz=50, Data (#hours)=101K, PT?=false2025.082.4033.59——0.711
CosyVoice2#Hz=25, Data (#hours)=167K, PT?=true2025.082.894.1——0.66
F5-TTS#Hz=-, Data (#hours)=101K, PT?=false2025.083.0153.508——0.629
Vevo2#Hz=12.5, Data (#hours)=101K, 7K, PT?=true2025.083.6393.814——0.693
Vevo2-base#Hz=12.5, Data (#hours)=101K, 7K, PT?=false2025.085.773.763——0.686
Vevo2-base#Hz=12.5, Data (#hours)=101K, PT?=false2025.085.8463.749——0.687
Genhancer confidence-based filteringFiltering threshold=top 70%, # Utterances=1960912026.0118.143.763.15——
Genhancer confidence-based filteringFiltering threshold=top 60%, # Utterances=1680782026.0118.783.733.12——
Genhancer confidence-based filteringFiltering threshold=top 80%, # Utterances=2241042026.0118.793.83.17——
Genhancer confidence-based filteringFiltering threshold=top 90%, # Utterances=2521172026.0119.293.763.14——
Genhancer confidence-based filteringFiltering threshold=top 50%, # Utterances=1400652026.0120.183.683.11——
Enhanced (unfiltered)# Utterances=280130, Data curation=Enhanced but unfiltered data2026.0120.453.643.1——
Source (noisy)# Utterances=280130, Data curation=Noisy source data2026.0121.312.732.74——
Vevo2-base#Hz=12.5, Data (#hours)=7K, PT?=false2025.0842.153.26——0.581