GUIDE

AI voice for lip sync animation

Lip sync is where a short animation eats the most time: every line needs a recording, and every recording needs mouth shapes timed to it. When the script changes, both have to be done again.

With the free SuperClonica engine the voice comes from text, and the free, open-source Rhubarb Lip Sync turns that voice into timed mouth shapes for Moho, OpenToonz, After Effects, Spine and other tools. This guide shows the two commands that connect them and how to keep iterating on the script without re-recording anything.

Browse voices Install the SuperClonica engine

From script to talking mouth

Because you already have the exact script, Rhubarb can use it as a hint, which makes the mouth shapes noticeably more reliable.

  • SuperClonica voices each line of the script as a WAV file.
  • Rhubarb Lip Sync reads WAV or Ogg Vorbis and writes the mouth shapes with their timing as TSV, XML, JSON, or DAT for Moho and OpenToonz.
  • Rhubarb lists integrations with After Effects, Moho, OpenToonz, Spine, Vegas Pro and Visionaire Studio; Blender users rely on community add-ons.
  • The voices speak 31 languages; Chinese is not supported yet.

Pick voices free online

On the page of any voice, type a line of your character, 30 to 100 characters, choose the language and press Voice it. In a few seconds you hear whether the voice fits the character, with no account.

Voice and analyse a line

Keep one line per file and give the text, the WAV and the mouth data the same name, so the animation tool always picks up the right pair.

  • Install the engine: python -m pip install superclonica-cpu.
  • Voice the line from its text file: superclonica --text-file line01.txt --voice M2 --lang en --output line01.wav.
  • Let Rhubarb read the same text as dialog and analyse the recording: rhubarb -d line01.txt -o line01.tsv line01.wav.
  • For lines that are not in English, use the language-independent recognizer: rhubarb -r phonetic -d line01.txt -o line01.tsv line01.wav. The default recognizer is for English only.

Change the script, not the session

When a line changes, edit its text file and run the two commands again: the new voice and the new mouth timing arrive together, so an animatic can change daily without a recording session. A fixed --seed keeps the same take for unchanged lines, --speed between 0.7 and 2.0 makes a line fit its shot, and --normalize evens the loudness between scenes.

Voices for animation you publish

For a film, series or game you publish, use the ten stock voices of the model, M1 to M5 and F1 to F5, your own voice trained in the SuperClonica studio from a recording of an actor who agreed, or a custom voice ordered on the exchange. Fan-made character voices in the catalog are for fan animation and parody, and are not endorsed by the rights holders.