Noise in Generated Audio !

[RifatMamayusupov]

I trained StableTTS model for my langauge. There is a problem that there is noise in generated audio. How to solve this issue:

this is result:

audio_link: https://drive.google.com/file/d/1zGdicK8qRIcue6Q5vCllPz7CvBnbQYm2/view?usp=sharing

@KdaiP Can you tell me what is problem and solve this , and which part of code should I focus to improve my model ?

[Popcorns333]

Change the inference starting point to a version with less amount of Gaussian noise.