List your company

Text-to-Speech on Japanese ambiguous pronunciation dataset (Female)

94.9AccuracyTKTO
47.99660.17372.3584.527Oct 7, 2025

Evaluation results

MethodLinks
TKTOPO Data=Unpaired2025.1094.97.52.9
KTOPO Data=Unpaired2025.1093.37.93
gpt-4o-mini-tts2025.109010.97.9
gemini-2.5-pro-preview-tts2025.1087.112.79.4
gemini-2.5-flash-preview-tts2025.1077.61410.5
DPOPO Data=Paired2025.1070.6127.6
Base model2025.1068.312.89
TKTOPO Data=Paired2025.1068.15.92.5
Supervised Fine-Tuning (SFT)PO Data=Desirable2025.1067.411.97.6
KTOPO Data=Paired2025.1065.46.62.8
F5-TTS with G2P2025.105013.610
F5-TTS2025.1049.817.318.9