Korokokok is a desktop studio that runs every step on your own hardware — AI-generated images, character-cast narration, and original soundtracks. No cloud, no subscription, no telemetry.
Windows 10/11 or macOS (Apple Silicon) · no account needed
Five workspaces share one project file. Write the story, build a cast, generate images, voice every character, score the chapter, and export.
Write it yourself or scaffold with a local LLM. Chapter-based, distraction-free.

Character-consistent AI imagery via local diffusion — Z-Image, Klein 4B, Qwen-Edit. Multi-angle character refs, prompts versioning, scene-aware composition.

Cast every character with a different voice. Audio drama mode bakes dialog overlays onto the narrator track. Five engines: Kokoro, Chatterbox, Orpheus, Qwen3-TTS, OmniVoice.

Three parallel soundtrack interpretations per chapter. Original songs with cover art and lyric videos. Reference-track-driven shaping.

Every retained asset — images, videos, prompts, narration, music — collected in one tabbed gallery. One click exports the whole chapter as a portable HTML folder you can hand off, share, or archive.

Your story, your voice samples, your generated assets — never leave your machine.
No subscription. No streaming credits. Free updates until the next major version.
Built to work on an 8 GB NVIDIA GPU on Windows, scaling to 16 GB+ for full quality. Also runs on macOS (Apple Silicon) — developed and tested on a Mac Mini M1.