The AI Music Arms Race: On the Detection of AI-Generated Music
Several companies now offer platforms for users to create music at unprecedented scales by textual prompting. As the quality of this music rises, concern grows about how to differentiate AI‑generated music from human‑made music, with implications for content identification, copyright enforcement, an...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Ubiquity Press
2025-06-01
|
| Series: | Transactions of the International Society for Music Information Retrieval |
| Subjects: | |
| Online Access: | https://account.transactions.ismir.net/index.php/up-j-tismir/article/view/254 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849716687556837376 |
|---|---|
| author | Laura Cros Vila Bob L. T. Sturm Luca Casini David Dalmazzo |
| author_facet | Laura Cros Vila Bob L. T. Sturm Luca Casini David Dalmazzo |
| author_sort | Laura Cros Vila |
| collection | DOAJ |
| description | Several companies now offer platforms for users to create music at unprecedented scales by textual prompting. As the quality of this music rises, concern grows about how to differentiate AI‑generated music from human‑made music, with implications for content identification, copyright enforcement, and music recommendation systems. This article explores the detection of AI‑generated music by assembling and studying a large dataset of music audio recordings (30,000 full tracks totaling 1,770 h, 33 m, and 31 s in duration), of which 10,000 are from the Million Song Dataset (Bertin‑Mahieux et al., 2011) and 20,000 are generated and released by users of two popular AI music platforms: Suno and Udio. We build and evaluate several AI music detectors operating on Contrastive Language–Audio Pretraining embeddings of the music audio, then compare them to a commercial baseline system as well as an open‑source one. We applied various audio transformations to see their impacts on detector performance and found that the commercial baseline system is easily fooled by simply resampling audio to 22.05 kHz. We argue that careful consideration needs to be given to the experimental design underlying work in this area, as well as the very definition of ‘AI music.’ We release all our code at https://github.com/lcrosvila/ai-music-detection. |
| format | Article |
| id | doaj-art-7aaa25230f814a06bb6e4090b725a665 |
| institution | DOAJ |
| issn | 2514-3298 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | Ubiquity Press |
| record_format | Article |
| series | Transactions of the International Society for Music Information Retrieval |
| spelling | doaj-art-7aaa25230f814a06bb6e4090b725a6652025-08-20T03:12:54ZengUbiquity PressTransactions of the International Society for Music Information Retrieval2514-32982025-06-0181179–194179–19410.5334/tismir.254254The AI Music Arms Race: On the Detection of AI-Generated MusicLaura Cros Vila0https://orcid.org/0000-0003-1098-6873Bob L. T. Sturm1https://orcid.org/0000-0003-2549-6367Luca Casini2https://orcid.org/0000-0002-3468-6974David Dalmazzo3https://orcid.org/0000-0002-3262-4091KTH Royal Institute of Technology, StockholmKTH Royal Institute of Technology, StockholmKTH Royal Institute of Technology, StockholmKTH Royal Institute of Technology, StockholmSeveral companies now offer platforms for users to create music at unprecedented scales by textual prompting. As the quality of this music rises, concern grows about how to differentiate AI‑generated music from human‑made music, with implications for content identification, copyright enforcement, and music recommendation systems. This article explores the detection of AI‑generated music by assembling and studying a large dataset of music audio recordings (30,000 full tracks totaling 1,770 h, 33 m, and 31 s in duration), of which 10,000 are from the Million Song Dataset (Bertin‑Mahieux et al., 2011) and 20,000 are generated and released by users of two popular AI music platforms: Suno and Udio. We build and evaluate several AI music detectors operating on Contrastive Language–Audio Pretraining embeddings of the music audio, then compare them to a commercial baseline system as well as an open‑source one. We applied various audio transformations to see their impacts on detector performance and found that the commercial baseline system is easily fooled by simply resampling audio to 22.05 kHz. We argue that careful consideration needs to be given to the experimental design underlying work in this area, as well as the very definition of ‘AI music.’ We release all our code at https://github.com/lcrosvila/ai-music-detection.https://account.transactions.ismir.net/index.php/up-j-tismir/article/view/254ai music detectionai musicgenerative aisunoudio |
| spellingShingle | Laura Cros Vila Bob L. T. Sturm Luca Casini David Dalmazzo The AI Music Arms Race: On the Detection of AI-Generated Music Transactions of the International Society for Music Information Retrieval ai music detection ai music generative ai suno udio |
| title | The AI Music Arms Race: On the Detection of AI-Generated Music |
| title_full | The AI Music Arms Race: On the Detection of AI-Generated Music |
| title_fullStr | The AI Music Arms Race: On the Detection of AI-Generated Music |
| title_full_unstemmed | The AI Music Arms Race: On the Detection of AI-Generated Music |
| title_short | The AI Music Arms Race: On the Detection of AI-Generated Music |
| title_sort | ai music arms race on the detection of ai generated music |
| topic | ai music detection ai music generative ai suno udio |
| url | https://account.transactions.ismir.net/index.php/up-j-tismir/article/view/254 |
| work_keys_str_mv | AT lauracrosvila theaimusicarmsraceonthedetectionofaigeneratedmusic AT bobltsturm theaimusicarmsraceonthedetectionofaigeneratedmusic AT lucacasini theaimusicarmsraceonthedetectionofaigeneratedmusic AT daviddalmazzo theaimusicarmsraceonthedetectionofaigeneratedmusic AT lauracrosvila aimusicarmsraceonthedetectionofaigeneratedmusic AT bobltsturm aimusicarmsraceonthedetectionofaigeneratedmusic AT lucacasini aimusicarmsraceonthedetectionofaigeneratedmusic AT daviddalmazzo aimusicarmsraceonthedetectionofaigeneratedmusic |