wtf is voice-pro?

Voice-Pro is a web application, built with the Gradio interface framework, that bundles several AI audio tools into one program. It runs locally on your own computer, mainly tested on Windows with an NVIDIA graphics card, and opens as a web page in your browser. Inside it you get speech recognition through Whisper and its faster variants, so you can turn spoken audio into text with timestamps. It has zero-shot voice cloning through tools called F5-TTS, E2-TTS, and CosyVoice, which means it can mimic a voice from a short sample without needing to train a custom model first. For generating speech from text it uses Edge-TTS and kokoro, with an optional Azure text-to-speech option if you bring your own Azure account keys. It can download YouTube videos and pull out just the audio, separate vocals from background music using a tool called Demucs, and translate text across more than 100 languages using Deep-Translator, again with an optional Azure Translator upgrade path. The README positions it as a free alternative to paid services like ElevenLabs, aimed at podcasters, video creators, researchers, and anyone doing multilingual dubbing or subtitling work. The project supports several languages in its own documentation, including Korean, English, Chinese, Japanese, German, Spanish, and Portuguese. One important note from the maintainers: active development has slowed because the team is focused on a separate project, so updates may be infrequent, though the code remains fully open source and free to use and modify. Setup uses a tool called uv for installing Python dependencies, and a troubleshooting tip in the README suggests deleting the installer_files folder and rerunning the start script if something breaks, since your downloaded AI models are kept separately and will not need to be redownloaded. The full README is longer than what was shown.