From Words to Waves: How Gender Bias Perpetuates AI-Based Music Generation

Bruce Wu

doi:10.54097/ykbyh440

Authors

Bruce Wu

DOI:

https://doi.org/10.54097/ykbyh440

Keywords:

AI-Generated Music, Media Bias, Acoustic Analysis.

Abstract

This study investigates the influence of gender bias in song lyrics on AI-generated music, focusing on how biases in textual content can impact various acoustic features in the resulting compositions. Using a dataset from the Billboard Hot 100, gender bias was quantified through a Transformer-based model, allowing for the categorization of songs into the most biased and least biased groups. These songs served as inputs for the Suno AI platform, which generated new music based on the provided lyrics and genres. Acoustic features such as Aggressiveness, Danceability, Approachability, Engagement, Valence, and Arousal were then analyzed to identify differences between the two groups. The results revealed significant disparities in several features, particularly in Valence and Arousal, indicating that gender bias in lyrics can influence AI-generated music’s emotional and rhythmic qualities. These findings demonstrate the potential for generative AI to perpetuate societal biases and highlight the importance of developing bias mitigation strategies. The study concludes by discussing theoretical and practical implications, proposing several methods to reduce bias in AI-generated content, and suggesting avenues for future Research to enhance fairness and inclusivity in AI-driven creative industries.

Downloads

Download data is not yet available.

References

[1] Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. Advances in neural information processing systems, 27, 2014.

[2] Mark S. Cole S. What is generative ai?, 2024. https://www.ibm.com/topics/generative-ai.

[3] Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.06125, 1(2):3, 2022.

[4] Open AI. Generative models,2016. https://openai.com/index/generative-models/.

[5] Bloomberg. Generative ai to become a $1.3 trillion market by 2032, research finds, 2017. https://www.bloomberg.com/company/press/generative-ai-to-become-a-1-3-trillion-market-by-2032-research-ffnds/.

[6] John Koetsier. Generative ai generation gap: 70% of gen z use it while gen x, boomers do not get it, forbes,2017. https://www.forbes.com/sites/johnkoetsier/2023/09/09/generative-ai-generation-gap-70-of-gen-z-useit-while-gen-x-boomers-dont-get-it/.

[7] Ed Lauder. Ai will power 95% of customer interactions by 2025, ai business, 2017. https://aibusiness.com/automation/ai-will-power-95-of-customer-interactions-by-2025.

[8] Aylin Caliskan, Joanna J Bryson, and Arvind Narayanan. Semantics derived automatically from language corpora contain human-like biases. Science, 356(6334):183–186, 2017.

[9] Tolga Bolukbasi, Kai-Wei Chang, James Y Zou, Venkatesh Saligrama, and Adam T Kalai. Man is to computer programmer as woman is to homemaker? debiasing word embeddings. Advances in neural information processing systems, 29, 2016.

[10] Saffya Umoja Noble. Algorithms of oppression: How search engines reinforce racism. In Algorithms of oppression. New York university press, 2018.

[11] Joy Buolamwini and Timnit Gebru. Gender shades: Intersectional accuracy disparities in commercial gender classiffcation. In Conference on fairness, accountability and transparency, pages 77–91. PMLR, 2018.

[12] H. Wiltshire. Revealing ai bias: Reinforcing harmful workplace stereotypes. rippl., 2024. https://rippl.work/blog/research-ai-bias-in-workplace-stereotypes/.

[13] Adi Robertson. Google apologizes for ‘missing the mark’ after gemini generated racially diverse nazis, 2024.https://www.theverge.com/2024/2/21/24079371/google-ai-gemini-generative-inaccurate-historical.

[14] Douglas Guilbeault, Sole`ne Delecourt, Tasker Hull, Bhargav Srinivasa Desikan, Mark Chu, and Ethan Nadler.Online images amplify gender bias. Nature, 626(8001):1049–1055, 2024.

[15] Ruha Benjamin. Race after technology: Abolitionist tools for the new Jim code. John Wiley & Sons, 2019.

[16] Yueyue Zhu, Jared Baca, Banafsheh Rekabdar, and Reza Rawassizadeh. A survey of ai music generation tools and models. arXiv preprint arXiv:2308.12982, 2023.

[17] Anna M Gorska and Dariusz Jemielniak. The invisible women: uncovering gender bias in ai-generated images of professionals. Feminist Media Studies, 23(8):4370–4375, 2023.

[18] Federico Bianchi, Pratyusha Kalluri, Esin Durmus, Faisal Ladhak, Myra Cheng, Debora Nozza, Tatsunori Hashimoto, Dan Jurafsky, James Zou, and Aylin Caliskan. Easily accessible text-to-image generation ampliffes demographic stereotypes at large scale. In Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency, pages 1493–1504, 2023.

[19] Jordan Vice, Naveed Akhtar, Richard Hartley, and Ajmal Mian. Quantifying bias in text-to-image generative models. arXiv preprint arXiv:2312.13053, 2023.

[20] Mi Zhou, Vibhanshu Abhishek, Timothy Derdenger, Jaymo Kim, and Kannan Srinivasan. Bias in generative ai. arxiv. arXiv preprint arXiv:2403.02726, 2024.

[21] Xiao Fang, Shangkun Che, Minjia Mao, Hongzhe Zhang, Ming Zhao, and Xiaohang Zhao. Bias of ai-generated content: an examination of news produced by large language models. Scientiffc Reports,14(1):5224, 2024.

[22] Jochen Hartmann, Jasper Schwenzow, and Maximilian Witte. The political ideology of conversational ai: Converging evidence on chatgpt’s pro-environmental, left-libertarian orientation. arXiv preprint arXiv:2301.01768, 2023.

[23] John J Hanna, Abdi D Wakene, Christoph U Lehmann, and Richard J Medford. Assessing racial and ethnic bias in text generation for healthcare-related tasks by chatgpt1. MedRxiv, 2023.

[24] Shicheng Xu, Danyang Hou, Liang Pang, Jingcheng Deng, Jun Xu, Huawei Shen, and Xueqi Cheng. Invisible relevance bias: Text-image retrieval models prefer ai-generated images. In Proceedings of the 47th international ACM SIGIR conference on research and development in information retrieval, pages 208–217, 2024.

[25] Sourojit Ghosh and Aylin Caliskan. ’person’== light-skinned, western man, and sexualization of women of color: Stereotypes in stable diffusion. arXiv preprint arXiv:2310.19981, 2023.

[26] David Rozado. The political biases of chatgpt. Social Sciences, 12(3):148, 2023.

[27] Fabio Motoki, Valdemar Pinho Neto, and Victor Rodrigues. More human than human: measuring chatgpt political bias. Public Choice, 198(1):3–23, 2024.

[28] Yixin Wan, Arjun Subramonian, Anaelia Ovalle, Zongyu Lin, Ashima Suvarna, Christina Chance, Hritik Bansal, Rebecca Pattichis, and Kai-Wei Chang. Survey of bias in text-to-image generation: Deffnition, evaluation, and mitigation. arXiv preprint arXiv:2404.01030, 2024.

[29] Ranjita Naik and Besmira Nushi. Social biases through the text-to-image generation lens. In Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society, pages 786–808, 2023.

[30] Yan Tao, Olga Viberg, Ryan S Baker, and Rene´ F Kizilcec. Cultural bias and cultural alignment of large language models. arXiv preprint arXiv, 2311, 2024.

[31] Hao-Wen Dong, Ke Chen, Julian McAuley, and Taylor Berg-Kirkpatrick. Muspy: A toolkit for symbolic music generation. arXiv preprint arXiv:2008.01951, 2020.

[32] John Biles et al. Genjam: A genetic algorithm for generating jazz solos. In ICMC, volume 94, pages 131–137.Ann Arbor, MI, 1994.

[33] Shaina Raza, Deepak John Reji, and Chen Ding. Dbias: detecting biases and ensuring fairness in news articles. International Journal of Data Science and Analytics, 17(1):39–59, 2024.

[34] Timo Spinde, Lada Rudnitckaia, Kanishka Sinha, Felix Hamborg, Bela Gipp, and Karsten Donnay. Mbic–a media bias annotation dataset including annotator characteristics. arXiv preprint arXiv:2105.11910, 2021.

[35] Dmitry Bogdanov, Nicolas Wack, Emilia Go´mez Gutie´rrez, Sankalp Gulati, Herrera Boyer, Oscar Mayor, Gerard Roma Trepat, Justin Salamon, Jose´ Ricardo Zapata Gonza´lez, Xavier Serra, et al. Essentia: An audio analysis library for music information retrieval. In Britto A, Gouyon F, Dixon S, editors. 14th Conference of the International Society for Music Information Retrieval (ISMIR); 2013 Nov 4-8; Curitiba, Brazil.[place unknown]: ISMIR; 2013. p. 493-8. International Society for Music Information Retrieval (ISMIR), 2013.

[36] [36] Anna Alajanki, Yi-Hsuan Yang, and Mohammad Soleymani. Benchmarking music emotion recognition systems. PloS one, pages 835–838, 2016.

[37] Jordi Pons and Xavier Serra. musicnn: Pre-trained convolutional neural networks for music audio tagging. arXiv preprint arXiv:1909.06654, 2019.

[38] Pablo Alonso-Jime´nez, Xavier Serra, and Dmitry Bogdanov. Music representation learning based on editorial metadata from discogs. In Rao P, Murthy H, Srinivasamurthy A, Bittner R, Caro Repetto R, Goto M, Serra X, Miron M, editors. Proceedings of the 23nd International Society for Music Information Retrieval Conference (ISMIR 2022); 2022 Dec 4-8; Bengaluru, India.[Canada]: International Society for Music Information Retrieval; 2022. p. 825-33. International Society for Music Information Retrieval (ISMIR), 2022.

[39] Cyril Laurier, Owen Meyers, Joan Serra, Martin Blech, and Perfecto Herrera. Music mood annotator design and integration. In 2009 Seventh International Workshop on Content-Based Multimedia Indexing, pages 156–161. IEEE, 2009.

[40] UNICEF. (2022). Gender action plan, 2022-2025: A vision for lasting, transformative change. Retrieved from https://www.unicef.org