Text to speech in character voices: voice any line as your favourite character
A character voice generator turns your text into speech that sounds like a character from a game, a cartoon, an anime or a film. It is how people make fan dubs, memes, game mods, stream alerts and voice lines for their own projects.
This guide shows how to do it with the SuperClonica catalog of character voices for the Supertonic 3 text-to-speech model: find the character, voice a line for free right on the site, and then voice as many lines as you need offline on your own computer.
How a character voice generator works
Each voice in the catalog is a voice style for Supertonic 3, an open text-to-speech model: one small JSON file that tells the model how the character sounds. The model reads any text you give it in that voice, so the character can say lines they never said.
Supertonic 3 speaks 31 languages, and every character voice can speak all of them: a character recorded in English can also read your text in Spanish, German or Russian. Chinese is not supported yet.
Step 1: find the character
Open the voice catalog and pick a category, or type the name in the search. The search understands different spellings and transliterations, so Geralt and Геральт, or a name with a typo, find the same voice.
- Video game characters: heroes and villains from popular games.
- Cartoon characters: animated series and films.
- Anime characters.
- Movie and TV characters.
- Every voice page has a demo, a description by its author and the languages it speaks.
Step 2: voice a line for free online
On the page of any voice, type a phrase of 30 to 100 characters and press Voice it. The phrase is synthesized by the SuperClonica engine, and you can listen to it and download the recording. No account is needed for this test.
It is the quickest way to hear whether the voice suits your line before you use it in a project.
Step 3: voice unlimited lines on your computer
For longer texts or many lines, download the voice file: free voices with one button and without an account, paid ones after purchase in your library. Then install the free SuperClonica engine with python -m pip install superclonica-cpu and voice a line with one command: superclonica --text "Your line" --voice character.json --lang en --output line.wav.
On your own computer there are no character limits or queues. On a desktop 8-core processor we estimate the engine at 6.4–9.7 times faster than real time, so a 10-second line is ready in about 1–1.6 seconds, without a graphics card.
Free character voices
Several character voices are free: among them Gollum, Trevor Philips from GTA V, Ciri, Dark Willow, Billy Butcher and Valera the Gummy Bear. They are on the free voices page of the catalog and in our GitHub repository with audio samples.
How to make a line sound in character
The voice gives the sound; the text gives the character. A few habits make the result much closer to the original.
- Write the line the way the character talks: their words, their exclamations, their rhythm.
- Use punctuation for timing: commas and full stops make pauses, an exclamation mark adds energy.
- Keep lines short. Two or three sentences per file are easier to redo than a whole page.
- Try a different seed if a take sounds flat: each seed reads the same text a little differently.
- Raise or lower the speed for a calmer or a more excited delivery.
What you may and may not do
Character voices are fan-made and are not endorsed by the rights holders. Use them for parody, fan works and personal projects within the SuperClonica terms, and never to deceive anyone or pass the result off as the original actor. For commercial use, make sure you have the rights you need.
The character you need is not in the catalog?
Search for the name anyway. If nothing is found, the catalog offers two options: order the voice from an author, who trains it on a recording you provide, or leave your email and we will write once when such a voice appears.