dAIrector (🤖 + 📖)

dAIrector is an automated director which collaborates with humans storytellers.

Documentation

Go to https://korymath.github.io/dairector/

Set Up

# install homebrew
/usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"
# install and upgrade portaudio, swig, git, python3
brew install --upgrade portaudio swig git python3
# set up the python3 virtual environment
virtualenv -p python3 env
# activate the virtual environment
source env/bin/activate
# install requirements
pip install -r requirements.txt
# in case of an error with pyaudio, may need to point to brew intall directly
# see https://stackoverflow.com/questions/33513522/when-installing-pyaudio-pip-cannot-find-portaudio-h-in-usr-local-include
# for more information
# pip install --global-option='build_ext' --global-option='-I/Users/korymath/homebrew/Cellar/portaudio/19.6.0/include' --global-option='-L/Users/korymath/homebrew/Cellar/portaudio/19.6.0/lib' pyaudio
# get the trained model and example files
wget https://storage.googleapis.com/api-project-941639660937.appspot.com/dairector_pretrained_examples.zip
# unpack the files
unzip dairector_pretrained_examples.zip

Run

# first ensure that your environment is activated
source env/bin/activate
# example 1a, generate a new subgraph from the entire plotto conflict graph
python markovgenerator.py -t outputfile.json plottoconflicts.json
# example 1b, interactive story telling using the plot subgraph and tv tropes hints
python -W ignore storyteller.py outputfile.json tvtropes.json tvtropesmodel.bin plottomodel.bin

Interactive Beat Generation

The storyteller is interactive, it understands the following commands:

• next [cue_text]
• hint [cue_text]
• quit

next uses the vector model from plottomodel.bin to find the next story beat based on the given cue text.

hint uses the tvtropesmodel.bin to find an appropriate trope.

Install pocketsphinx

pip install pocketsphinx==0.1.15
pip install PyAudio==0.2.11

Basic Usage

Improvisors on stage can cue the system to provide the next plot point or the next hint. The improvisors provide the dialogue for each plot clause.

Training a new model

python topicvectors.py tvtropesmodel.bin tvtropes.json

Cite

@inproceedings{eger2018dairector,
  author = {{Eger}, M. and {Mathewson}, K.~W.},
  title = "{dAIrector: Automatic Story Beat Generation through Knowledge Synthesis}",
  booktitle = {AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE18), Joint Workshop on Intelligent Narrative Technologies and Intelligent Cinematography and Editing},
  publisher = {AAAI},
  year = 2018,
  address={Edmonton, Alberta, Canada},
  month = 10,
}

License

This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/3.0/ or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA.