An advanced singing voice synthesis system with high fidelity, expressiveness, controllability and flexibility based on DiffSinger: Singing Voice Synthesis via Shallow Diffusion Mechanism
Apache License 2.0
2.62k
stars
275
forks
source link
Add objective evaluation metrics for duration predictor #127
Rhythm Correctness: to evaluate the correctness of rhythm, i.e. word duration fitness.
Phoneme Duration Accuracy: to evaluate the fitness of phoneme durations after a rhythm alignment.
Rhythm Correctness: to evaluate the correctness of rhythm, i.e. word duration fitness. Phoneme Duration Accuracy: to evaluate the fitness of phoneme durations after a rhythm alignment.