This paper presents Text2Fart, a sophisticated AI model that utilizes advanced natural language processing and innovative sound synthesis techniques to generate highly realistic and amusing flatulence sounds based on textual inputs. Our model is based on a comprehensive corpus of fart sound assets that have been tagged with natural language descriptions, allowing for highly effective and precise sound synthesis. We have also employed mathematical modelling techniques to examine factors such as pitch, timbre, duration, resonance, and vibratory behavior, providing deep insights into the underlying principles of flatulence acoustics. Our adutio diffusion model incorporates these models into its training process, resulting in highly accurate and amusing fart sound synthesis. Through quantitative and qualitative evaluation, we demonstrate the effectiveness and novelty of Text2Fart in injecting fun and play into the ongoing discourse surrounding AI technology. Our research provides a tongue-in-cheek perspective on the potential of AI in the realm of humor and entertainment, highlighting the importance of a balanced diet of humor in the ever-evolving field of artificial intelligence.
@article{doo2020fartnet,
title={FartNet: Classifying Fart Sounds Using a Convolutional Neural Network},
author={Doo, Joe and Smelly, Jane and Pooter, John},
journal={International Journal of Flatulence Research},
volume={7},
number={2},
pages={42--49},
year={2020},
publisher={Flatulence Society}
}
@inproceedings{tootman2021gastrogan,
title={GastroGAN: Synthesizing Flatulence with Varying Degrees of Wetness using Generative Adversarial Networks},
author={Tootman, Tom and Gasper, Gary},
booktitle={Proceedings of the Conference on Flatulence and Gastrointestinal Disorders},
pages={88--92},
year={2021},
organization={ACM}
}
@article{odorferous2022stink2vec,
title={Stink2Vec: Representing Farts in a Continuous Vector Space},
author={Odorferous, Abigail and Foul, Fred and Whiff, Walter},
journal={Journal of Flatulence and Odor Science},
volume={8},
number={1},
pages={15--22},
year={2022},
publisher={Flatulence Society}
}