Skip to content

Releases: Eddycrack864/AI-Cover-Maker

AI Cover Maker v2.0.0

Choose a tag to compare

@Eddycrack864 Eddycrack864 released this 02 Nov 19:22

AI Cover Maker 2.0 is a from-scratch rewrite of RVC AI Cover Maker UI. It makes an AI cover in one click: the song is separated with the UVR models of python-audio-separator, the vocals are converted with RVC (Applio), and everything is mixed back together with vocal effects. It is built on UVR5 UI and runs on Python 3.12, Gradio 6, audio-separator 0.47 and PyTorch 2.7.1 (CUDA 12.8).

📦 Downloads

Package Download Size SHA256
Windows (precompiled) AI-Cover-Maker-v2.0.0-windows.zip 4.06 GB 60e87bae26a24acdf752cae15f3e6a25f009f88bbeb193bfa9a4076a4e240fb1
Linux (precompiled) AI-Cover-Maker-v2.0.0-linux.tar.gz 4.70 GB 1422386d61deb22de5469c555ba614059169df0cca1358f755c213db36ccb55e
Source code See the assets below, or git clone https://github.com/Eddycrack864/AI-Cover-Maker.git

Precompiled version (nothing to install, the Python environment is included):

  1. Download the package for your system.
  2. Extract it. It contains an AI-Cover-Maker folder (on Linux: tar -xzf AI-Cover-Maker-v2.0.0-linux.tar.gz).
  3. Run run-AI-Cover-Maker.bat (Windows) or ./run-AI-Cover-Maker.sh (Linux).

The Linux package needs glibc 2.35 or newer (Ubuntu 22.04, Debian 12, Fedora 36 or later). Both packages include PyTorch 2.7.1 with CUDA 12.8 (NVIDIA GTX 10 series or newer, RTX 50 series supported). FFmpeg is still needed, and Git to update.

Normal installation: clone the repository (or download the source code and rename the folder to AI-Cover-Maker), then run AI-Cover-Maker-installer.bat / AI-Cover-Maker-installer.sh and start the app with run-AI-Cover-Maker.bat / run-AI-Cover-Maker.sh. See the README for the prerequisites.

No GPU? Use the Colab.

⬆️ Coming from RVC AI Cover Maker UI

AI Cover Maker is a new project in a new repository, so it is a new installation (the old updater doesn't bring you here). Your voice models can come with you: copy each folder of the old logs/ folder (a .pth and its .index) to models/rvc/ of AI Cover Maker and press Refresh voice models, or add them again from the Voices tab.

✨ What's new

Cover

  • One click from song to cover: vocals/instrumental separation → optional lead/backing split, de-reverb, de-echo and denoise → RVC → vocal effects → mix.
  • Presets: Fast, Balanced and Best quality, or choose the model of every step yourself.
  • Each step only lists the models made for it, with the recommended one first and marked with a ⭐. Show all models lists the whole catalog.
  • Every step is cached: making the same song with another voice, effects or mix only runs what changed, so the second cover is much faster.
  • Backing vocals: convert them with a second voice, or keep the original singer.
  • Change the key of the whole song keeping its tempo.
  • Vocal effects: high-pass, compressor and reverb (AICoverGen's chain by default), and a mix that never clips.
  • Every stem is saved with the cover, in outputs/<song> (<voice> Cover)/: vocals, instrumental, lead/backing vocals, reverb, converted vocals... for your own remixes. Output as wav, flac, mp3, ogg or m4a.
  • Songs from a link (every site supported by yt-dlp) and batch covers.

Voice conversion

  • RVC inference from Applio: rmvpe, fcpe, crepe and swift; index ratio, protect, volume envelope, autotune, automatic pitch, formant shifting, custom embedders and noise cleanup.
  • Voices tab: add voice models by link (Hugging Face, Google Drive, direct .zip/.pth/.index) or upload them, and delete them.
  • On GPUs with 8 GB or less, the separation model and the voice model are never loaded at the same time.

Models

  • The full UVR5 UI catalog: 169 Roformer, MDX23C, MDX-NET and VR models, downloaded on first use from GitHub or the Hugging Face mirror, with resumable downloads.

App

  • Save settings: every option of the Cover tab is loaded every time the app starts.
  • 19 languages, themes, Discord Rich Presence and screen reader support, like UVR5 UI.
  • Installers and updaters for Windows and Linux with a private Python 3.12 environment, precompiled versions and a Colab notebook.
  • Docs with every setting explained, a troubleshooting guide and status checkers (info/).

Under the hood

  • The app is the ai_cover_maker/ package: a UI-independent core (separation, RVC, effects, mixer and the cached cover pipeline), the Gradio UI, and about 200 tests run on Windows and Linux by GitHub Actions.

🙌 Contributors

Thanks to ShiromiyaG, the original creator of RVC AI Cover Maker UI, whose features and ideas are the base of this rewrite, and to everyone who used the old app, reported issues and suggested features.