Generate
Your Voice signed IP
Save a named preset - a base voice plus your FX. Presets appear in the Voice dropdown. Every clip you generate is signed with an Ed25519 key stored on this device, tying the output to you.
How it works
Speak or type
Tap the mic and talk, or type your script. Up to 2,000 characters per generation.
Pick a voice
Fourteen neural voices across US, UK, and Australian accents. Or save your own preset.
Shape the sound
Pitch shift, speed control, gain, reverb, whisper mode, robot mode - real DSP on every clip.
Signed to you
Every clip is cryptographically signed with an Ed25519 key generated on your device. Your output, your IP.
Self-hosted engine
Runs on your own Hetzner server. No third-party API, no per-character billing.
Voice breeding (Lab)
Coming: interpolate speaker embeddings to grow brand-new voices that belong to nobody but us.