Mirrors/Retrieval-based-Voice-Conversion-WebUI
1
0
Fork 0
mirror of https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI.git synced 2026-08-29 10:09:32 +02:00
Code Issues Packages Projects Releases Wiki Activity

Update Changelog_EN.md

Browse Source
This commit is contained in:
RVC-Boss
2026-07-21 10:33:56 +08:00
committed by GitHub
parent 8f2fdbf483
commit cd3e966d3c
1 changed files with 13 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
docs/en/Changelog_EN.md
View File

@@ -1,3 +1,16 @@
### 2026-07-18
- Reduced web UI complexity (no path required for audio inference input, improved real-time frontend logs, automatic model-list refresh after saving a small inference model, automatic detection of the matching index path after selecting a model on the inference home page, etc.).
- Data feature extraction, model training, and index building can now be interrupted and resumed after restarting.
- Automatically detects the GPU to determine the inference mode and precision.
- Fixed dependency issues (including issues with pyworld, fairseq, matplotlib, PyAV, and ONNX Runtime GPU environments).
- Resolved encoding issues.
- The web UI now automatically searches for an available port at startup.
- Single-GPU training no longer uses DDP.
- The base model remains unchanged. Several issues affecting output quality have been fixed. Updating the bundled package/code and rerunning inference with an existing fine-tuned model, or retraining and then running inference with the new version, may improve results.
- Added support for the FCPE pitch extraction algorithm.
- Added support for CUDA Graph inference acceleration.
- Removed features that were clearly obsolete or no longer useful.
### 2023-10-06
- We have created a GUI for real-time voice change: go-realtime_gui.bat/realtime_gui.py (Note that you should choose the same type of input and output device, e.g. MME and MME).
- We trained a better pitch extract RMVPE model.