Text-to-Speech on IndicTTS Database Marathi
4.3MOSGround Truth
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Ground Truth2022.11 | 4.3 | — | — | 9.3 | |
| AI4Bharat-TTSAcoustic Model=FastPitch, Vocoder=HiFiGAN2022.11 | 3.26 | 9.51 | 0.3 | 7.5 | |
| Tacotron2+WaveGlowAcoustic Model=Tacotron2, Vocoder=WaveGlow2022.11 | 2.69 | 14.84 | 0.41 | 13.6 | |
| GlowTTS+HiFiGANAcoustic Model=GlowTTS, Vocoder=HiFiGAN2022.11 | 2.61 | 11.53 | 0.5 | 148.4 |