Applio

Applio wraps Retrieval-based Voice Conversion in a local Gradio app maintained by the AI Hispano community. You load or download an RVC voice model, feed in a vocal recording, and get the same performance sung or spoken in the target voice. The interface also trains new voice models from your own clean dataset with TensorBoard monitoring, and adds realtime conversion for a live microphone, text-to-speech routed through a voice model, a blender that merges two voice models, an audio analyzer, a model downloader and plugins. Install with the run-install script on Windows, Linux or macOS or a precompiled Windows zip, with Docker and Colab as alternatives; conversion works on CPU, training is far quicker on an NVIDIA GPU or a cloud notebook, and AMD on Windows has a ZLUDA guide. Code and bundled weights are MIT, so commercial use is allowed, though the unmodified official build also carries a terms-of-use file requiring lawful, consented use. Maintainers call it mature and now ship mostly security and dependency updates, with occasional features. Suited to cover makers and developers who want RVC without assembling the toolchain.