The problem
10+ AI sessions in parallel. Each one calling ElevenLabs. $100 a month, quota burned in days, tabs speaking over each other.
Open sourceFree forever
We hit a wall—cost, complexity, vendor lock. So we built the fix and open-sourced it. Each block does one thing, runs anywhere, and costs nothing.
First release
LOCAL VOICE CLONE / XTTS V2
Clone your voice. Speak anything. Pay nothing.
Record 30 seconds of yourself talking. golden-voice clones your voice locally using XTTS v2. No cloud, no API keys, no monthly bill. One command and you’re cloned.
Full details + install guide# one-time setup
golden-voice setup my-voice.wav
# speak in your own voice
golden-voice speak "Ship it."
# pipe anything
cat summary.md | golden-voice speak --stdin The origin
golden-voice wasn’t planned in a roadmap. It appeared when an expensive, fragile workflow finally became impossible to ignore.
10+ AI sessions in parallel. Each one calling ElevenLabs. $100 a month, quota burned in days, tabs speaking over each other.
macOS say + sox for instant recaps. XTTS v2 for voice cloning. Five rounds of dependency fixes. By evening, it worked.
We wrapped the pain—dependency pins, torch patches, codec installs—so nobody else has to fight it.
The philosophy
GOLDEN FOCUS / WORKING NOTE
Every block exists because we got burned. Then we fixed it for everyone.
No freemium. No “contact sales.” Open source, and staying that way.
One thing, done well. A clean interface that runs anywhere.
Repair the world piece by piece. Not by preaching—by shipping.
The next blockStarts with friction
Every block begins with a problem worth removing. See what we’re fixing now, or bring us the next scar.