Voice01
How a character sounds, and when it speaks.
Explore voice
Speech synthesis turns language into sound; rhythm, pitch, emphasis and pauses shape delivery. Listening, turn-taking and interruption handling matter as much as voice quality.
Livelier Memories Generative LabsHuman-created.
Human-created.
Generatively powered.
The applied generative practice behind character-led experiences.
We integrate and evaluate third-party generative technology around our original characters, while developing our own character systems, workflows and local capabilities.
Consistent character identity.
Flexible technology underneath.
Creative identity stays distinct from the technology delivering it. Systems can evolve without redefining the character; integration, review and testing still matter.
Six layers. One character.
These scientific concepts and design questions inform our direction. They are not a list of features available in every experience.
Language02
A distinct personality. A relevant response.
Explore language
Context and approved knowledge help ground responses; persona guidance defines tone, role and boundaries. These support consistency, not infallibility, and the character should acknowledge uncertainty.
Movement03
Performance beyond the words.
Explore movement
Lip-sync aligns visible mouth movement with speech. Expressions, gaze and head movement shape performance. The challenge is coordinating face and voice without excessive motion.
Image04
New scenes. A recognisable character.
Explore image
References and structural controls can guide appearance, pose and layout. Art direction and review check identity drift and unwanted changes across new scenes.
Story05
Continuity gives imagination meaning.
Explore story
Motivation, structure and consequences make a story. Planning, context and revision help continuity; human editorial judgment decides which generated possibilities serve the character.
Harness06
The parts have to work together.
Explore harness
Our harness concept brings character identity, approved knowledge, behaviour rules and coordination between language, voice and visuals together. Evaluation asks whether the experience is useful, consistent and within its boundaries, not merely convincing.
Here today. Looking ahead.
Today
Our current work integrates third-party generative technology for live video and text character experiences, alongside approved knowledge and a curated visual collection. Human review and version control support how content is prepared and updated.
Research directions
Development directions — not completed results or promised features.
- Our own character systems, workflows and local capabilities
- More consistent identity across formats
- Smoother listening, speaking and animation
- Longer-story continuity
- Systematic behaviour evaluation
- Clearer authorship and permissions records
Human judgment belongs in the system.
Creative work deserves clear attribution and permission. Character experiences need appropriate privacy practices, candid disclosures and ways to handle failure. These are requirements to design for, not a promise that generative systems never get things wrong.
Begin with a character.
Meet the personalities at the heart of Livelier Memories.
Choose your characterExplore Character & Mascot Science