Amphion (/æmˈfaɪən/) is a toolkit for Audio, Music, and Speech Generation. Its purpose is to support reproducible research and help junior researchers and engineers get started in the field of audio, music, and speech generation research and development.
I have noticed that the current project uses Harvest as the F0 predictor algorithm. To improve the robustness and accuracy of pitch detection, I suggest adding RMVPE (Robust Multi-View Pitch Estimation) as an alternative F0 predictor.
Advantages of RMVPE
Robustness: RMVPE combines the strengths of multiple pitch detection algorithms, making it more resilient to noise and unstable audio signals.
Accuracy: By integrating outputs from various algorithms, RMVPE can enhance the accuracy of pitch estimation, especially in complex audio environments.
Implementation Details
I am not entirely sure if this suggestion is the best fit for the project, and I welcome any feedback and discussion from the community.
Thank you for your valuable advice!
In fact, we have already integrated RMVPE into Amphion in development and will update it to the public repository after further testing, so stay tuned!
Suggestion to Add RMVPE as an F0 Predictor
I have noticed that the current project uses Harvest as the F0 predictor algorithm. To improve the robustness and accuracy of pitch detection, I suggest adding RMVPE (Robust Multi-View Pitch Estimation) as an alternative F0 predictor.
Advantages of RMVPE
Implementation Details
I am not entirely sure if this suggestion is the best fit for the project, and I welcome any feedback and discussion from the community.
Looking forward to hearing your thoughts on this.