Computer-Generated Music for Tabletop Role-Playing Games

This repository contains the source code to reproduce the results of the AIIDE'20 paper Computer-Generated Music for Tabletop Role-Playing Games. This paper presents Bardo Composer, a system to generate background music for tabletop role-playing games. Bardo Composer uses a speech recognition system to translate player speech into text, which is classified according to a model of emotion. Bardo Composer then uses Stochastic Bi-Objective Beam Search, a variant of Stochastic Beam Search that we introduce in this paper, with a neural model to generate musical pieces conveying the desired emotion.

Examples of Generated Pieces

Installing Dependencies

Reproducing Results

Bardo Composer uses a fine-tuned BERT to classify the story emotion and a fine-tuned GPT2 to classify the music emotion. In the paper, we report the accuracy of these models. This section describes how to reproduce the results we found.

Story Emotion Classification

We compared the fine-tuned BERT model for story emotion classification with the simpler Naıve Bayes approach of [Padovani, Ferreira, and Lelis (2017)].

Fine-tune Pre-trained BERT

cd composer/clf_dnd/
python3 clf_bert.py --data ../../data/dnd/

Train Naive Bayes

cd composer/clf_dnd/
python3 clf_nbayes.py --data ../../data/dnd/

Both these scripts will perform and report accuracy experiments on the Call of the Wild dataset [Padovani, Ferreira, and Lelis (2017)].

Music Emotion Classification

We compared the fine-tuned GPT2 model for music emotion classification with the simpler LSTM approach of [Ferreira and Whitehead (2019)].

Download Pre-trained GPT-2

In the paper, the GPT-2 model was pre-trained using a new dataset called ADL-Piano-Midi. The pre-trained model can be download as follows:

$ wget https://github.com/lucasnfe/bardo-composer/releases/download/0.1/pre-trained-gpt2.zip

Fine-tune Pre-trained GPT-2

cd composer/clf_vgmidi/
python3 clf_gpt2.py --conf clf_gpt2.json

Train LSTM

cd composer/clf_vgmidi/
python3 clf_lstm.py --conf clf_lstm.json

Citing this Work

If you use this method in your research, please cite:

@inproceedings{ferreira2020computer,
  title={Computer-generated music for tabletop role-playing games},
  author={Ferreira, Lucas and Lelis, Levi and Whitehead, Jim},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment},
  volume={16},
  number={1},
  pages={59--65},
  year={2020}
}