Text to speech voice for VTubers
Not every VTuber wants to stream with their real voice: some protect their privacy, some cannot talk for hours, some stream in a language they read better than they speak. A text-to-speech voice lets the avatar speak whatever you type, in a voice that belongs to the character.
SuperClonica gives your character a natural voice, free to test online and running on your own computer during the stream. This guide shows how to choose the voice, how to speak typed lines live and how to keep the voice yours.
Who a TTS voice is for
A TTS voice is not a voice changer: it speaks what you type, not what you say into a microphone.
- Streamers who prefer not to use their real voice on stream.
- Creators who stream in a language they type better than they speak.
- Characters that need the same voice every stream, without a microphone setup.
- Channels that mix live talk with a TTS companion or narrator.
Find the voice free online
On the page of any voice, type a typical line of your character, 30 to 100 characters, choose the language and press Voice it. In a few seconds you hear whether it fits your avatar, with no account.
Speak typed lines live
Keep lines short; two quick sentences sound more natural on stream than one long one.
- Run the engine as a local service: docker run -d -p 8000:8000 -v superclonica-models:/models superclonica/superclonica:cpu.
- A small script or hotkey sends each typed line to POST /synthesize and plays the WAV it returns.
- A short phrase takes well under a second on an ordinary processor, so the delay feels like a pause before speaking.
- Voice catchphrases and greetings in advance and put them on stream-deck buttons.
Make the avatar talk
Send the TTS audio through a virtual audio cable that your avatar software listens to as a microphone, so the mouth moves with the voice, and add the same audio to your streaming software. Keep one voice and one seed for the character, so it sounds the same every stream. Tell viewers that the character speaks with a synthetic voice; most audiences like knowing how the channel works.
A voice that is really yours
For a monetized channel, use a voice you have the rights to. The ten stock voices of the model, M1 to M5 and F1 to F5, are free; for a unique character voice, train one in the SuperClonica studio from a recording of a voice actor who agreed, on a computer with an NVIDIA graphics card, or order one on the exchange. Character and parody voices from the catalog are fan-made: keep them for jokes and fan content, not as your persona’s permanent voice.