RVC-Project / RVC-Project/Retrieval-based-Voice-Conversion-WebUI
Minor output alignment issue
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 38.4k
- Forks
- 5.3k
- PR merge metrics
- No merged PRs in 30d
Description
When there are more than one segments in vc.pipeline, every new segment makes subsequent audio output out of place by one frame (i.e. length of self.window = 0.01s). The final audio is thereby also slightly shorter. This effect is barely perceptible, but can be verified using audio editing software.
After experimenting with several attempted fixes, this effect can be drastically reduced after changing line 388, 405, 423 and 440 of infer/modules/vc/pipeline.py from "self.t_pad_tgt : -self.t_pad_tgt" to "(self.t_pad_tgt - self.window) : -(self.t_pad_tgt - self.window)."
I was able to cut the input vs output length difference of a 4-minute audio from 0.04 seconds down to 0.002 seconds.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start in infer/modules/vc/pipeline.py at lines 388, 405, 423, and 440, where segment padding is applied. Reproduce the issue with multiple segments in vc.pipeline and compare input and output lengths using audio editing software. Done means the output alignment improves and the length difference is reduced without introducing a new issue.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- audio-video-rtc
- Issue type
- Bug
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Stale
- Clarity
- Clearly specified
- Newbie friendliness
- 58/100