Home
Welcome to the public release of the ARIA-VALUSPA Platform (AVP). The platform is developed as part of ARIA-VALUSPA, a EU Horizon 2020 project running from 1 January 2015 to 31 December 2017. The work was funded by European Union Horizon 2020 research and innovation programme, grant agreement No 645378. The AVP has been adopted by the Nottingham Biomedical Research Centre, which will maintain it until April 2022.
The current version (3.0) was released on 23 February 2018.
With AVP, you can build your own Virtual Humans. They can be run in Ogre3D, or in Unity3D, the latter also allowing Augmented and Virtual Reality settings. Two example videos can be found here:
- Alice: https://youtu.be/XaiIqPJdCiw
- Dr Jones: https://youtu.be/Zb4fouzgY6M
Our Virtual Human Technology is provided in a modular fashion, with in particular distinct Behaviour Analysis, Dialogue Management and Behaviour Generation modules. See the figure below for a schematic of the framework.
The Framework comes with the following features:
- Automatic Speech recognition (ASR, in Kaldi)
- Paralinguistic analysis (OpenSMILE)
- Written in JSON/Java-script compliant Flipper 2.0
- Outputs FML to Behaviour Generation block
- GRETA for visual behaviour generation
- CereProc's CereVoice for Text to Speech (high quality voices available separately from CereProc)
- Unity3D port to put an AVP character in your own Game/AR/VR environment
This wiki contains information how to install the Framework, detailed documentation you need when adapting the framework to your own needs, and some tutorials to get you started.