Talking Head Generation
Definition
Talking portraits are created as videos of a speaking face from image, audio, or text inputs.
The generation of talking portraits produces video sequences of a speaking face from image, audio, or text inputs.
Simply explained
From a portrait and suitable inputs, a moving representation is created in which the face appears to speak.
Why it matters in practice
The term helps to select, develop, evaluate AI systems in a well-founded manner and realistically assess their technical risks.
Example from business
A training video uses an explicitly authorized fictional character and labels the synthetic generation.
Limits and pitfalls
- Representing a lip-sync method as full portrait generation or ignoring consent, identity deception, provenance labeling, artifacts, and culturally inappropriate gestures.
Origin & history
Research on facial animation, audio control, and lip synchronization has become more realistic through deep generative models. Wav2Lip is specifically a method for audiovisual lip synchronization, not the overarching term for the task.
Use in mid-sized companies
- Possible application: A training video uses an explicitly authorized fictional character and labels the synthetic generation.
Frequently asked questions
What does the generation of talking portraits mean?
The generation of talking portraits produces video sequences of a speaking face from image, audio, or text inputs. What mistakes should be avoided? Representing a lip-sync method as full portrait generation or ignoring consent, identity deception, provenance labeling, artifacts, and culturally inappropriate gestures.
Sources
- A Lip Sync Expert Is All You Need for Speech to Lip Generation In The Wild · Prajwal et al., A Lip Sync Expert Is All You Need for Speech to Lip Generation In The Wild
Source checked on 10.10.2026. Editorial use cases are hypothetical; no evidence of real customer results.
Editorially reviewed on: