cl-voice clones a voice from a short sample and generates TTS in it. cl-audio does local stem separation — pull vocals from any track, or make a music-only bed. Both wire straight into the editor via the desktop bridge.
The AI agent talks to your local desktop app to make edits. Install it once — you can skip this if it's already running.
Pick the AI you use. The command installs the "cl-voice" skill so your agent knows how to control this part of the app.
copilot skill add https://raw.githubusercontent.com/Shubham-vish/contentlead/main/skills/cl-voice/SKILL.mdgit clone https://github.com/Shubham-vish/contentlead.git && copilot skill add ./contentlead/skillsOpen your AI in the terminal (or your IDE), then paste any of these prompts to try it.
No. cl-voice routes through the desktop bridge which uses your signed-in session cookie. Voice cloning happens on ContentLead's servers.
10–30 seconds of clean speech in a single voice works well. Longer isn't better — quality of the sample matters more than length.
cl-audio is fully local. No files leave your machine. Uses demucs (or the shipped model) under the hood.