Software & data
Tools and resources
Software
MultiMod platform
Acquisition platform for synchronised multimodal speech data: motion capture, electromagnetic articulography, audio and video. Used for the corpora behind most of our synthesis work, and open to external collaborations.
VisArtico
Visualisation of articulatory data acquired by electromagnetic articulography (EMA): coil trajectories, vocal-tract views, synchronised audio and annotation.
Talking head animation system
Speech-driven facial animation, from acoustic and textual input to a 3D character. The research line that led to Dynalips.
Patents
-
2021
Audio-driven speech animation using recurrent neural network
S. Ouni, T. Biasutto-Lervat, S. Dahmani. International patent WO2021023861, filed 11 February 2021. The coarticulation model behind it came out of Théo Biasutto-Lervat's thesis, and led to the creation of the Dynalips spin-off in September 2023.
-
2018
Image processing device
S. Ouni, G. Gris. International patent US2018/0061109 A1. 2018
Data
Through our Multimod platform, we have conducted numerous multimodal data acquisitions, both for our own research projects and for other laboratories and researchers. These acquisitions include articulatory data using EMA, audiovisual data using motion capture (MoCap), and gestural data, also captured using MoCap.