Applicability of WaveDiff for 1D Time-Series Signal Synthesis

[zyt0002]

Hi WaveDiff team,

I'm exploring WaveDiff for a 1D time-series signal synthesis task, specifically for generating ECG signals from PPG inputs(biomedical direction) . I understand WaveDiff is designed for waveform generation, but I would like to clarify its applicability to 1D time-series tasks.
Here are some questions:
1.Has WaveDiff been tested or adapted for 1D time-series signal synthesis tasks, such as PPG-to-ECG conversion or similar medical signal generation?
2.Are there any specific modifications (e.g., model architecture, preprocessing) required to apply WaveDiff to 1D time-series data instead of audio waveforms?
3.Are there any available examples or pretrained models for 1D signal synthesis tasks that I could reference?
4.How does WaveDiff compare to other generative models (e.g., Diffusion Models, Flow Matching) in terms of performance and efficiency for 1D signal synthesis?

[zyt0002]

Looking forward to your reply!