MotionVox technology and development
Qualiteg develops Expression Aware® for facial animation and Voice Wrapper Technology® for vocal expression. Together, they power realistic avatars that explain your content with natural voices and expressions.
See how this technology supports training, product presentations and public information, alongside custom avatars and deployment choices for your organisation.
Qualiteg works across the development process: sourcing and preparing licensed data, training AI on our GPU cluster, building the infrastructure that generates videos, and delivering the application.
Contracts and usage permission from performers
Organise and clean the training data
Train models on our GPU cluster
Build the environment for video generation
Deliver video creation and editing tools
This development capability supports improvements to expressions and voices, as well as organisation-specific avatars and deployment options.
Create presenter-led videos without arranging filming or narration sessions. Edit the script and regenerate the video when product information or training content changes.
Expression Aware®
Our avatar engine recreates lip movements for Japanese speech, subtle expressions and natural nods. In the editor, place emotes to time expressions and reactions to the script.
Explore expression and motion editingVoice Wrapper Technology®
We also develop our own voice-expression technology. AI voices based on recordings by professional voice actors let you choose an avatar and voice for each video. Staff can create training or product presentations without appearing or recording their own voices.
Explore avatars and voicesGenerate multiple language versions with the same avatar and voice. Reduce the need to arrange presenters and narrators for each language when creating product videos, staff training or visitor information. Translation also takes the pronunciation of proper names into account.
A MotionVox template example
A menu introduction combining food photos and an avatar. Play the Japanese, English and Chinese versions right here.
Edited and generated entirely in MotionVox. Replace the food photos, dish names, prices and script to introduce your own restaurant.
Template ID 6B9SN9 Enter this ID in MotionVox’s template search to find the same template.
We consider the permissions behind faces and voices alongside the quality of the generated video, so organisations can plan ongoing use.
Developing our avatar technology and inference infrastructure allows us to offer a Japan-hosted processing plan and on-premises configurations in your own environment.
A plan that completes processing in Japan without sending input data to overseas servers. Discuss its suitability when your organisation has requirements about where scripts and documents are processed.
Generate videos within your own environment. We can discuss a configuration based on the information you handle, the number of users and your production volume.
Japan-hosted and on-premises options are arranged individually. Available features, including translation, and the scope of processing depend on the configuration. Please share your use case and information-management requirements.
Our research team publishes technical explanations of speech-to-mouth modelling, audio-video synchronisation and stable model training—the engineering behind avatar generation.
Explains how speech features relate to mouth shapes, including the challenge of accounting for surrounding sounds.
Discusses conventional and machine-learning approaches to correcting timing drift that can accumulate in longer videos.
Introduces averaging model weights during training. MotionVox also uses EMA to smooth changes in model parameters.
See the expressions and voices in a demo, discuss an avatar in your uniform, or explore processing internal documents in Japan. We can walk you through the options for your use case.
Contact us / Free demo