This experimental production will be a daily TV show based on music clips, with weather information complementary section. The working title is "Hack the Van". It is aimed at young people. All the sections will be presented by virtual characters. The program will be automatically generated, so it will be produced without any person.
The main character, Sefi, presents the sections. She talks and move through a randomized actions and pre-recorded voice sentences (to check if it is possible to randomize movements and facial gestures depending of the voice or of semantic tags).
The broadcasted music clips are chosen by viewers by SMS voting, and a robot called Kilo presents the clips, reading the titles from a data base, with a synthetic voice (TTS). This section is aimed to check if a TTS voice is usable on a TV context, depending of the character who use it and of the script. It also wants to test the RT generation of voice and movements reading of a dynamic Data Base.
A weather man called Sam makes the weather forecast for main cities. This section experiments with emotional tags on an English TTS system based on a pre-recorded corpus.
This First Prototype consists of a short demonstration program with the first scenes of the main presenters and the forecast section. Main effort of first prototype has been addressed to the technical achievements. Second Prototype will have an artistic enhancement.
The automatic animation platform has generated two versions of the programme. In each version the role of the characters is changed, in order to test the movements reusement and the voice transformation.
The program is 100% automatically generated from the automatic animation platform, based on a integration of several SALERO tools.
Reuse of movements: Automatic adaptation of the same movement in all production characters (independently of the character proportions)
Automatic random scripts: Every generation of the program changes the script; Enriched automation capabilities.
External playlist data retrieval: "Presenter 2" character introduces the music videos reading the title from a data base in real time; automation and interactivity enhancement.
Template-based animation: Video generation based on a XML template (from the program structure)
Combination of different audio generation sources: Combination of TTS, recorded voice and high-quality TTS (technical and artistic test) and use of voice transformation.