Rolecast
self-hosted multi-character roleplay voice studio
speech
chatterbox
kokoro
minimax
phonon
qwen3-tts
ollama
Scene
What happens
Tone (optional)
Characters
+ add character
Turns
Writer LLM
Auto (MiniMax M3 first)
MiniMax M3 (cloud)
Ollama (self-hosted)
Voice engine
Auto (best available)
MiniMax Speech (332 voices, ~2s/line)
Chatterbox (expressive, 28 voices)
Phonon 31M (0.4s/line, 37 voices)
Qwen3-TTS 0.6B (premium local, 9 voices)
Kokoro (fast, 72 voices)
Language
Generate scene
ready.
Result
No scene generated yet.