The Virtual Storyteller is a multi-agent framework for automatic story
generation in two stages: Simulation (the content of the story
is created by means of a multi-agent system) and
Presentation (the story content is
expressed in natural language and told by a speaking embodied agent).
The Virtual Storyteller is a multi-agent framework for automatic story
generation. It creates new, original stories based on the
actions of autonomous agents who live in a simulated story world. A
formal story representation is constructed on the basis of the
"adventures" of these agents and then expressed in natural language
(text and speech). We are also working on story visualisation. Below
we discuss the main components of the system.
Simulation (content generation)
This is a multi-agent system responsible for
generating story plots. The agents involved in the simulation are:
Character Agents. These agents inhabit the Story World and
can perform actions there. They have individual emotions, beliefs,
and can make plans to achieve their goals.
World Agent. This agent is responsible for keeping track of
the current state of the Story World. It processes actions from the
Character Agents and updates the world accordingly.
Plot Agent. Intermediates between the Character Agents and
the World Agent, and creates a formal representation of the story
content called the fabula. The fabula forms the basis for the next
stage of story generation: the presentation.
Presentation
The Narrator is responsible for converting the
formal content representation (the fabula) to a natural language text.
Tasks include mapping story elements to sentences, combining these
sentences using proper cue words ("because", "but", etc.), choosing
suitable referring expressions ("Diana", "the beautiful princess",
"she"), and ensuring proper morphology and word order.
The story can be presented to the user in the form of text, but it can
also be actually told to the user. This requires converting the
text of the story to speech that is suitable for storytelling.
Storyteller speech is much more expressive than the neutral speech
produced by standard text-to-speech systems. Therefore rules have been
designed to convert standard synthesized speech to storytelling
speech.
Listen to the difference between the following synthetic
speech samples:
Example 1 (neutral sentence): standard
- storytelling
style
Example 2 (suspenseful):
standard
-
storytelling
style
Note that these rules are not yet used in the actual presentation by
the embodied storyteller, for which we use DEIRA, the
winner of the 2007 GALA
Award for Lifelike Agents. You can watch a movie of DEIRA telling
a story here:
Website
The Virtual Storyteller website, with documents and publications on
different aspects
of the Virtual Storyteller, can be found here:
http://wwwhome.cs.utwente.nl/~theune/VS/
WIKI/blog pages
Here you can find descriptions of various
aspects of the Virtual Storyteller, such as the authoring process
(with detailed examples):
Students interested in working on the Virtual Storyteller for a Capita
Selecta, research project or Masters thesis, or others who would like
to know more about the Virtual Storyteller, please contact Mariët Theune.
M. Theune, S. Rensen, H.J.A. op den Akker, D.K.J. Heylen and A. NijholtEmotional characters for automatic plot creation, in TIDSE 2004: Technologies for Interactive Digital Storytelling and Entertainment, S. Göbel, U. Spierling, A. Hoffmann, I. Iurgel, O. Schneider, J. Dechau and A. Feix (eds), Lecture Notes in Computer Science, volume 3105, Springer Verlag, Berlin, ISBN 3-540-22283-9, ISSN 0302-9743, pp. 95-100, 2004 [ BiBTeX ] []
I.M.T. Swartjes and M. TheuneA Fabula Model for Emergent Narrative, in Technologies for Interactive Digital Storytelling and Entertainment, Third International Conference, TIDSE 2006, S. Göbel, R. Malkewitz and I. Iurgel (eds), Lecture Notes in Computer Science, volume 4326, Springer Verlag, Heidelberg, ISBN 978-3-540-49934-3, pp. 49-60, 2006 [ BiBTeX ] []
M. Theune, N. Slabbers and F. Hielkema The Narrator: NLG for digital storytelling, in ENLG-07 11th European Workshop on Natural Language Generation, S. Busemann (eds), DFKI Document Series, DFKI (Deutsches Forschungszentrum für Künstliche Intelligenz GmbH), Germany, ISSN 0946-0098, pp. 109-112, 2007 [ BiBTeX ] []
M. Theune, N. Slabbers and F. Hielkema The automatic generation of narratives, in Proceedings of the 17th Meeting of Computational Linguistics in the Netherlands (CLIN17), P. Dirix, I. Schuurman, V. Vandeghinste and F. Van Eynde (eds), LOT Occasional Series, LOT, Utrecht, ISBN 978-90-78328-41-4, pp. 131-146, 2007 [ BiBTeX ] []
I.M.T. Swartjes, E.E. Kruizinga, M. Theune and D.K.J. HeylenEmergent Narrative and Late Commitment, in Proceedings of the Eight International Conference on Intelligent Virtual Agents, H. Prendinger, J. Lester and M. Ishizuka (eds), Lecture Notes in Artificial Intelligence, volume 5208/2008, Springer Verlag, Berlin, ISSN 0302-9743, pp. 543-544, 2008 [ BiBTeX ] []
S. Louchart, I.M.T. Swartjes, M. Kriegel and R.S. Aylett Purposeful authoring for emergent narrative, in Proceedings of the First Joint International Conference on Interactive Digital Storytelling, Springer Verlag, Berlin, ISBN 978-3-540-89424-7, pp. 273-284, 2008 [ BiBTeX ] []
I.M.T. Swartjes and M. TheuneAn Experiment in Improvised Interactive Drama, in Intelligent Technologies for Interactive Entertainment, INTETAIN 2009, Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering, volume 9, Springer Verlag, Berlin, ISBN 978-3-642-02314-9, ISSN 1867-8211, pp. 234-239, 2009 [ BiBTeX ] []