This is a speech-to-speech voice transformer that runs ElevenLabs models through the inference.sh CLI. You feed it an audio file and pick from 22+ voices (British George, American Aria, Australian Charlie, etc.), and it changes the speaker while keeping the words and emotional tone intact. Works in 70+ languages with the multilingual model. The obvious use case is creating character voices from a single voice actor, but it's also handy for anonymizing recordings or upgrading rough narration to something more polished without re-recording. You can chain it with their voice isolator to clean noisy audio first, or their dubbing tool if you need translation too. Requires the belt CLI and an inference.sh account.
npx skills add https://github.com/inferen-sh/skills --skill elevenlabs-voice-changer