Skip to content

Release Vevo1.5#426

Merged
RMSnow merged 12 commits intoopen-mmlab:mainfrom
RMSnow:vevosing
Apr 12, 2025
Merged

Release Vevo1.5#426
RMSnow merged 12 commits intoopen-mmlab:mainfrom
RMSnow:vevosing

Conversation

@RMSnow
Copy link
Collaborator

@RMSnow RMSnow commented Apr 12, 2025

✨ Description

  1. We release Vevo1.5, which extends Vevo and focuses on unified and controllable generation for both speech and singing voice. The pre-trained models and training code are both included. See this blog for more details.
  2. We release the training code of Vevo.

🚧 Related Issues

#400

Copy link

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copilot reviewed 54 out of 69 changed files in this pull request and generated no comments.

Files not reviewed (15)
  • egs/codec/coco/contentstyle_fvq16384_12.5hz.json: Language not supported
  • egs/codec/coco/contentstyle_fvq16384_12.5hz.sh: Language not supported
  • egs/codec/coco/prosody_fvq512_6.25hz.json: Language not supported
  • egs/codec/coco/prosody_fvq512_6.25hz.sh: Language not supported
  • egs/codec/vevo/fvq32.json: Language not supported
  • egs/codec/vevo/fvq32.sh: Language not supported
  • egs/codec/vevo/fvq8192.json: Language not supported
  • egs/codec/vevo/fvq8192.sh: Language not supported
  • egs/svc/AutoregressiveTransformer/ar_synthesis.json: Language not supported
  • egs/svc/AutoregressiveTransformer/ar_synthesis.sh: Language not supported
  • egs/svc/FlowMatchingTransformer/fm_contentstyle.json: Language not supported
  • egs/svc/FlowMatchingTransformer/fm_contentstyle.sh: Language not supported
  • egs/vc/AutoregressiveTransformer/ar_conversion.json: Language not supported
  • egs/vc/AutoregressiveTransformer/ar_conversion.sh: Language not supported
  • egs/vc/AutoregressiveTransformer/ar_synthesis.json: Language not supported
Comments suppressed due to low confidence (2)

bins/vc/train.py:57

  • The '--resume' argument is defined as a boolean flag, but the help text indicates it expects a model name. Please update the help text to clarify that this flag resumes training from a checkpoint.
parser.add_argument("--resume", action="store_true", help="The model name to restore")

bins/codec/train.py:59

  • The '--resume' argument is a boolean flag, yet the help text suggests it provides a model name. Please revise the help message for consistency with its boolean behavior.
parser.add_argument("--resume", action="store_true", help="The model name to restore")

Copy link
Collaborator Author

@RMSnow RMSnow left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Self reviewed.

Copy link
Collaborator

@jiaqili3 jiaqili3 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lgtm

Copy link
Collaborator

@viewfinder-annn viewfinder-annn left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lgtm

@RMSnow RMSnow merged commit 1ebed0c into open-mmlab:main Apr 12, 2025
1 check passed
KhryptorGraphics pushed a commit to KhryptorGraphics/Amphion that referenced this pull request Feb 2, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants