Home // AIMEDIA 2025, The First International Conference on AI-based Media Innovation // View article

Deepfake Music and Listener Sentiments: A Large-Scale Analysis of YouTube Comments.

Authors:
Francisco Tigre Moura
Visieu Lac

Keywords: Artificial Intelligence; Sentiment analysis; Deepfake; Deepfake Music.

Abstract:
This study investigates emotional responses to deepfake music by analyzing 31,363 YouTube comments using 'twitter-roberta-base-sentiment-latest' model. The research addresses an important gap in the literature by focusing on user sentiments toward AI (Artificial Intelligence)-generated deepfake songs that mimic human voices, contrasting them with non-deepfake AI music (instrumental or relaxation genres). Findings reveal a surprising predominance of positive and neutral sentiments toward deepfake music, particularly in genres like Rock/Pop and Cartoon, though negative reactions are more pronounced when renowned human voices are mimicked. The study also identifies genre-specific patterns and a longitudinal decline in novelty-driven engagement, especially within Rap/Hip-Hop. Compared to non-deepfake AI music, which consistently triggers positive sentiments, deepfake music evokes more mixed responses, suggesting that voice mimicry remains a critical aspect. The findings contribute to the understanding of how AI creativity influences listener emotions and perceptions, while raising timely questions regarding authenticity and acceptance of deepfake art.

Pages: 1 to 6

Copyright: Copyright (c) IARIA, 2025

Publication date: July 6, 2025

Published in: conference

ISBN: 978-1-68558-330-9

Location: Venice, Italy

Dates: from July 6, 2025 to July 10, 2025