Voice Synthesis for Animation
Voice synthesis for animation is the use of AI text-to-speech and voice-cloning models to produce spoken performance, used in animation production for placeholder voice during animatic stages, localisation across markets, and, on lower-stakes work, final delivery.
Voice synthesis is currently the most production-ready of the AI generation techniques. Modern models produce voice that is hard to distinguish from a real recording for short forms and conversational tone. The result is good enough for many production uses, though hero brand voiceover is still recorded with real performers.
Inside our pipeline, voice synthesis appears at three points. At animatic stage, a synthesised voice unblocks pre-production: the script is locked, the voice is generated, the animatic moves on without waiting for a recording session. At localisation, voice cloning ports a brand performance across many languages. At delivery, on lower-stakes work, synthesised voice can replace a recording, with full rights cleared.
On work like Inchstones, the hero brand performance is recorded with real performers because the brand voice carries the campaign. Synthesised voice would not be appropriate as final. For internal versions, animatic placeholders, and language ports, the AI side of the pipeline does the work.
Myth Labs handles voice synthesis production work, with a particular focus on video localisation with AI for global brand campaigns.
Related
Related services
Sources
Academic papers, recognised industry standards, and canonical industry texts that back up claims in this entry.
- Capture, Learning, and Synthesis of 3D Speaking Styles. Siarohin, A. et al., arXiv, 2019Supports: voice operated character animation
- VOCAL: Vowel and Consonant Layering for Expressive Animator-Centric Singing Animation. Liang, L. et al., SIGGRAPH, 2023Supports: audio-driven facial animation
Frequently asked questions
Is synthesised voice better than a real performer?
Not for hero work. A trained performer brings craft, range, and direction-following that synthesis still cannot match. For hero brand voiceover, we record. For animatic placeholder, video localisation, and lower-stakes work, synthesis is part of the stack.
Can a brand clone its existing voice talent?
Yes, with consent and a licence. We commission performers under terms that allow voice cloning where the campaign needs it, agreed up front. Without a licence, we will not clone.
Are there union and rights questions?
Yes, and they vary by region. The UK Equity, US SAG-AFTRA, and other unions have evolving positions on AI voice. We track the contract terrain project by project and only ship work where the rights are clean.