Seperated steps for DiffSinger #1487
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
oxygen-dioxide
previously approved these changes
Apr 11, 2025
yqzhishen
requested changes
Apr 11, 2025
Collaborator
yqzhishen
left a comment
yqzhishen
left a comment
There was a problem hiding this comment.
I would argue a default 10 steps for pitch model inference. Maybe 5 is a little too radical.
Contributor
|
Ok! |
yqzhishen
approved these changes
Apr 12, 2025
KakaruHayate
added a commit
to KakaruHayate/OpenUtau
that referenced
this pull request
Apr 20, 2025
Seperated steps for DiffSinger stakira#1487
stakira
approved these changes
May 7, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Added ability to set different step values for DiffSinger models: Acoustic, Variance and Pitch
This will speed up the rendering and give more options.
According to my experience, only 5 steps are enough to render a good Pitch, while Variance parameters require a value around Acoustic, but can still be much lower if it doesn't contain complex parameters. So I suggest leaving it initially the same as Acoustic, so that the user can change it on their own if they see fit.