Derpy-Turtle-The-Kokoro-Trainer Hatches Smooth Voice Clones Locally

Derpy-Turtle-The-Kokoro-Trainer is a Windows GUI that blends Kokoro’s text-to-speech with RVC voice conversion to build better local voice clones. It lets you search for and refine Kokoro voice tensors, train a voice conversion model on your own reference audio, and automatically generate a final converted speech file. The whole pipeline runs inside a single launcher that bootstraps its own Python environment.

BovineOverlord created the project after discovering that chasing a high optimizer score alone didn’t deliver convincing matches. He designed a two‑stage method where Kokoro handles clear, stable pronunciation and RVC applies the target timbre. The trainer runs entirely on a local Windows machine and wraps complex steps into one executable.

A complete voice cloning pipeline in one window

The tool suits prosumer GPU owners, privacy‑conscious professionals, and indie creators who need local voice cloning. It helps them produce voiceovers, game dialogue, or narration without cloud reliance or complex scripting. The guided GUI and presets make it accessible even to non‑developers.

Lessons from building a local voice cloner

BovineOverlord found that long random‑walk searches alone rarely delivered natural‑sounding results, which led him to add the RVC stage. He stresses using clean audio—10 to 30 minutes minimum, free of reverb, music, or background chatter—because data quality matters more than quantity. The project is free for non‑commercial use; anyone interested in commercial licensing needs to reach out directly.

"The practical goal is simple: generate clear Kokoro speech, then use a target-trained RVC model to move the final audio closer to the desired voice." — Source: GitHub