Three Creative Engines
Every Voice, Every Style, Every Scene
Three purpose-built synthesis engines for every production scenario — from scripted narration to zero-shot character creation.
Command Pre-Trained Personas
Drive a library of professionally-tuned vocal characters using plain English style instructions. Consistent, high-quality output across every take — perfect for narration, audiobooks, and reliable character voice-over.
- 9 built-in personas — Ryan, Aiden, Vivian, Serena, Eric, Dylan, Sohee, and more
- Style Injection: "Speak softly", "Conspiratorial whisper", "Old radio announcer"
- Style & Profile Manager to create, save, and toggle custom styles inline
- Seed control for perfectly reproducible takes
Create Voices That Never Existed
Generate entirely new vocal identities from text descriptions alone. Define the body and the performance. The model constructs a unique vocal fingerprint from scratch — no reference audio required.
- Two-field formula: Voice Description (the body) + Style Instruction (the performance)
- "A 60-year-old gravelly smoker with a Southern drawl" — just describe it
- Seed control to lock and reproduce exact vocal fingerprints
- Export to Voice Clone for consistent cross-session rendering
Precision Digital Replicas
Capture any voice from as little as 3–10 seconds of reference audio. Feed it any script and the model delivers that same voice, performing your direction. The integrated Prep Station handles reference transcription automatically.
- Only 3–10 seconds of clean reference audio required
- Integrated Whisper AI transcription for reference preparation
- Speaker prompt caching — computed once per batch, not per block
- || delimiter for multi-segment long-form content rendering
