Pith. sign in

REVIEW 3 major objections 4 minor 31 references

Assessing AI-generated music detection in real-world broadcast monitoring

T0 review · 3 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper introduces BAMM, a 40-hour real-TV dataset, and finds that current CNN detectors for AI-generated music drop from near-perfect $F_1$ scores on clean tracks to 0.186 (clean-trained) and 0.472 (broadcast-trained) on real…

desk verdict Useful real-world benchmark dataset, but the RTB detection numbers are entangled with the label-generation pipeline, so the headline degradation is plausible but the specific magnitudes are not established. read the letter →

arxiv 2608.07359 v1 pith:LXIWS4V3 submitted 2026-08-07 eess.AS cs.AI

David López-Ayala , Fernando García de la Cruz , Pablo Zinemanas , Emilio Molina , Martín Rocamora This is my paper · ORCID
classification eess.AScs.AI
keywords AI-generatedmusicdetectionbroadcastmonitoringBAMMdatasetCNNdetectordomaingapaudiofingerprintingreal-worldevaluationAIartifacts
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that current CNN-based AI-generated-music detectors, which look close to perfect on isolated clean tracks, are not reliable when deployed on real television broadcasts. To test this, it introduces BAMM, a 40-hour dataset of actual TV recordings labeled as AI-generated or human-made through audio fingerprinting and a strict five-model consensus. On this dataset, the clean-trained CNN drops from $F_1$ 0.992 under controlled conditions to $F_1$ 0.186, and the broadcast-trained variant drops to $F_1$ 0.472, with both models showing substantial score overlap between AI and human music. The finding matters because broadcast monitoring is the setting where AI-music detection would actually be used, and synthetic broadcast benchmarks substantially overstate how ready these tools are.

What carries the argument

The load-bearing machinery is the BAMM curation pipeline plus a fixed CNN architecture. BAMM is built by taking clean reference tracks, labeling the AI class only when five different detectors—including the model later evaluated as CNN Clean—agree unanimously, finding their occurrences in a global TV archive through landmark audio fingerprinting, and filtering retrieved clips with a deep music detector that keeps only foreground or background music. The evaluated models are six-layer CNNs on mel-spectrograms that are identical except for training data: CNN Clean uses clean foreground music only, while CNN Broadcast mixes the same music with speech at random signal-to-noise ratios and encodes it to 8 kHz mono at 40 kbps to emulate broadcast degradation. Because the architecture is fixed, the comparison isolates the effect of training domain, and BAMM supplies the first real-TV test set against which the synthetic broadcast benchmark can be checked.

What would settle it

Independently re-label a random sample of BAMM clips using official release provenance, artist metadata, or expert listening, and compare with the ensemble labels; if independent labels disagree often on the AI class, the reported RTB $F_1$ drop is inflated by selection bias, while strong agreement would confirm that current detectors genuinely fail on real broadcasts.

Watch

Extended reading notes

Core claim

The central discovery is that the domain gap between clean foreground music and real TV broadcast audio is large enough to break current artifact-based CNN detectors. In the paper's three-scenario evaluation, both the clean-trained and broadcast-trained variants achieve $F_1$ above 0.99 on clean foreground music (CFM), but on synthetic broadcast mixtures (STB) they fall to $F_1$ 0.342 and 0.661, and on the real broadcast clips of BAMM (RTB) they fall to $F_1$ 0.186 and 0.472, with ROC-AUC 0.707 and 0.775 respectively. The score distributions show the failure mode: human-made clips are usually scored near zero, but many AI-generated clips also receive low scores, so the models miss AI music far more often than they false-alarm on human music. The authors interpret this as evidence that broadcast masking, low-bitrate encoding, and background placement hide the decoding artifacts these models were trained on, making current training approaches insufficient for reliable detection in broadcast monitoring.

Load-bearing premise

All BAMM AI labels come from unanimous agreement among five clean-trained detectors, so the reported real-broadcast degradation could partly reflect that particular labeling rule rather than the intrinsic difficulty of AI-generated music in TV audio.

Editorial extensions

If this is right

  • Neither clean-trained nor broadcast-trained CNNs can currently serve as a reliable AI-generated-music monitor on real TV: best-case $F_1$ is 0.472 with AUC 0.775 on RTB.
  • Adding speech mixtures and low-bitrate encoding to training helps—CNN Broadcast beats CNN Clean on every broadcast scenario—but the remaining gap shows that training-domain matching alone does not solve the problem.
  • Synthetic broadcast benchmarks overstate real-world readiness: STB $F_1$ scores (0.342 and 0.661) are consistently higher than RTB $F_1$ scores (0.186 and 0.472).
  • Background music is harder than foreground music for both models, with foreground AUC values of 0.782 and 0.858 versus background AUC values of 0.667 and 0.745 on BAMM.
  • The dominant failure is missed AI content rather than false alarms, since human clips concentrate near zero while many AI clips are also scored low.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the AI label requires unanimous agreement among five clean-trained detectors, BAMM's AI class likely contains only tracks that are easy to flag in clean conditions; if so, recall on arbitrary AI broadcast music could be even lower than the reported $F_1$ values.
  • A natural next experiment, not run here, is to fine-tune or augment the detector on BAMM clips and test on held-out channels and dates; large gains would show the gap is distributional, while small gains would point to an intrinsic loss of detectable artifacts in 8 kHz/40 kbps audio.
  • The 8 kHz mono, AAC-LC proxy stream removes much of the high-frequency spectrum where decoding artifacts live, so detectors using different front-ends—for example source-separation or artifact fingerprints—may behave quite differently from these mel-spectrogram CNNs on the same BAMM clips.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper introduces BAMM, a 40-hour dataset of real television broadcasts containing AI-generated and human-made music, and uses it to evaluate two CNN-based detectors: CNN Clean, trained on clean foreground music, and CNN Broadcast, trained on synthetic broadcast mixtures. The models are compared across three scenarios: Clean Foreground Music (CFM), Synthetic TV Broadcast (STB), and Real TV Broadcast (RTB). The authors report near-perfect CFM performance, substantial degradation in STB, and further degradation in RTB, with CNN Broadcast outperforming CNN Clean but still showing limited F1 and AUC. The paper concludes that current CNN training approaches are insufficient for reliable AI-generated music detection in real-world broadcast monitoring.

Significance. If the RTB evaluation were valid, the paper would provide a valuable real-world benchmark and a sobering result: state-of-the-art CNN detectors that perform near-perfectly on clean audio degrade dramatically on actual broadcast recordings. The public release of BAMM, the baseline code, and the evaluation scripts are concrete contributions that could enable future work on robust detection and domain adaptation. The paper's main finding—that synthetic broadcast degradation does not fully capture the difficulty of real broadcast conditions—is plausible and important. However, the RTB evaluation is compromised by the label-selection procedure, and the specific F1/AUC magnitudes in Table 1 cannot be interpreted as unbiased estimates of real-world performance.

major comments (3)
  1. [Section 3.1, Table 1] The AI-labeled reference tracks in BAMM are selected by requiring unanimous agreement from a five-detector ensemble, and the first detector listed is CNN Clean, the very model whose RTB performance is reported in Table 1. Consequently, the RTB positive class is restricted to tracks that CNN Clean already recognizes under clean foreground conditions. This selection can bias the measured F1/AUC in either direction, but in any case the RTB numbers do not measure performance on a representative sample of AI-generated music in broadcast. The paper should either hold CNN Clean out of the labeling ensemble, construct an independent label set (e.g., via metadata or manual expert annotation), or at minimum report results on a subset of BAMM labeled without CNN Clean's involvement, so that the headline degradation is not an artifact of the evaluation protocol.
  2. [Section 4, Conclusions] The benchmark explicitly focuses on Suno v3.5, and the paper acknowledges that model-agnostic detection remains open. However, the abstract and conclusions state broadly that 'current training approaches on CNN-based detectors remain insufficient for reliable AI-generated music detection in broadcast monitoring.' Since only a single generator family/version is evaluated, this claim is too strong. The conclusions should be qualified to Suno v3.5, or the evaluation should include at least one additional generator (e.g., a newer Suno version or a different system) to support the general statement.
  3. [Section 3.2–3.3, Section 4.2.3] BAMM is built by fingerprint-matching clean reference tracks to broadcast occurrences and then applying a Deep Music Detector filter. The paper does not report how many candidate tracks failed at each pipeline stage, nor how representative the final 40-hour set is of all AI-generated music in the broadcast archive. Without a dropout analysis, it is unclear whether the RTB results reflect the difficulty of the broadcast domain or the peculiarities of the subset of tracks that have clean reference versions, are findable by fingerprinting, and survive the DMD filter. The paper should provide stage-by-stage counts and discuss potential selection effects on the measured performance.
minor comments (4)
  1. [Section 5, Table 1] The table reports F1 and ROC values without confidence intervals or any measure of uncertainty. Given that the paper's main numerical claims rest on differences such as F1 0.186 vs. 0.472, the authors should provide bootstrap confidence intervals or at least a significance test to show that the differences are not due to the specific choice of clips or thresholds.
  2. [Section 3.1, Figure 2] The temporal analysis in Figure 2 is presented as validation of the labeling strategy, but it only shows aggregate score distributions over time. It would be more informative to show that the unanimous-agreement subset is stable under perturbation of the ensemble composition, since the ensemble includes the model under evaluation.
  3. [Section 4.1.2] The description of CNN Broadcast training states that speech segments are concatenated to span the full music duration and mixed at random SNR between -30 dB and +30 dB. It is unclear whether the speech is mixed at constant gain or whether the SNR is drawn per clip; this detail should be clarified for reproducibility.
  4. [Section 2, references] The related work section cites several relevant papers, but the discussion of detection robustness would benefit from a reference to recent work on out-of-distribution generalization for audio deepfake detection, which is closely connected to the domain-gap argument made here.

Circularity Check

1 steps flagged · score 6.0 of 10

BAMM's AI labels are assigned by an ensemble that includes CNN Clean, the model evaluated in Table 1; the RTB numbers therefore measure a label-selected subset, not independent real-world performance.

  1. self definitional [Section 3.1 (Reference Track Selection), with RTB evaluation in Section 4.2.3 and Table 1]
    "For the AI-generated class, a candidate track was included only if it exceeded the calibrated decision thresholds across all five detectors, requiring unanimous agreement from the ensemble. ... The first model is a publicly available convolutional neural network presented and assessed in this paper under the name CNN Clean."

    BAMM's AI class is defined by requiring unanimous agreement of five detectors, one of which is CNN Clean, the model whose RTB F1/AUC is reported in Table 1. Consequently, every RTB AI test clip has a clean reference that CNN Clean already classified as AI above its calibrated threshold; CNN Clean's clean-domain false negatives cannot appear in the positive class. The RTB evaluation therefore measures CNN Clean on a subset selected by its own clean-domain decisions, not on a representative sample of AI-generated music in TV. The direction of the selection bias is unclear: it could favor easily detectable robust artifacts or codec-fragile ones, but either way the specific F1/AUC values are not an independent estimate of real-world performance.

full rationale

The central circular step is unambiguous: Section 3.1 states that a reference track is admitted to the AI class only if all five detectors, including CNN Clean, exceed calibrated thresholds. Section 4.2.3 and Table 1 then evaluate CNN Clean on the resulting BAMM RTB set. This makes the RTB positive class partially self-defined with respect to the model being benchmarked, so the reported RTB degradation is not an unbiased estimate for the general population of AI music in broadcast. The paper's broader conclusion that real-world broadcast is harder than synthetic benchmarks may still survive, because the STB scenario (AI-OpenBMAT) does not use CNN Clean to construct its labels, and even a cherry-picked positive subset can exhibit domain shift. However, the specific F1/AUC magnitudes and BAMM's validity as an independent reference benchmark are compromised. No other load-bearing circularity was found: the CFM and STB evaluations use external or previously published data, the architecture follows Afchar et al. [6], and the self-citation to [7] for AI-OpenBMAT is not itself used to define the evaluated models' outputs. The score of 6 reflects that one of the three headline evaluations reduces partly by construction, while the overall research direction retains independent content.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central evaluation rests on BAMM labels that are partly generated by the evaluated model's own clean-domain predictions, plus several unverified domain assumptions about fingerprinting, DMD, and the human-made baseline. No new theoretical or physical entities are introduced.

free parameters (5)
  • Global mel-spectrogram mean and standard deviation = mean = -4.0, std = 3.0
    Chosen constants for input standardization in Section 4.1; both models use this fixed normalization.
  • CNN Broadcast mixture ratio = 70% mixed / 30% clean
    Ratio of speech-mixed to clean samples in broadcast-oriented training, selected by the authors in Section 4.1.2.
  • SNR range for speech-music mixing = -30 dB to +30 dB
    Uniform random SNR for simulated broadcast training examples in Section 4.1.2.
  • Ensemble calibration thresholds = not reported
    Per-model decision thresholds set to zero false positive rate on Da-TACOS plus a 500-track AI control set in Section 3.1; values are not disclosed.
  • Fingerprinting and DMD matching parameters = not reported
    Private implementations used for BAMM construction; parameters and accuracy are not given in Sections 3.2 and 3.3.
assumptions (5)
  • domain assumption Audio fingerprinting correctly identifies broadcast occurrences of reference tracks
    BAMM construction relies on a private landmark-based fingerprinting implementation in Section 3.2 to align clean references to TV occurrences, with no reported accuracy figures.
  • ad hoc to paper Unanimous agreement of the five-detector ensemble is a valid proxy for AI generation
    Section 3.1 defines the AI class by unanimous detector agreement rather than by ground-truth provenance, embedding detector behavior into the labels.
  • domain assumption Tracks released between 2020 and 2022 are human-made
    Section 3.1 restricts the human-made class to releases predating Suno v3.5, assuming no comparable public AI music existed then; this is not verified by provenance or listening.
  • domain assumption DMD music and speech labels are accurate
    Sections 3.3 and 5.3 depend on the Deep Music Detector for filtering and foreground or background analysis, but no validation accuracy is reported.
  • ad hoc to paper Suno v3.5 is representative of AI-generated music in broadcast
    The benchmark and dataset focus on Suno v3.5 as stated in Section 4, so generalization to other generators is not established.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Assessing AI-generated music detection in real-world broadcast monitoring." pith.science (2026). https://pith.science/paper/LXIWS4V3

@misc{pith2026260807359,
  author       = {Pith},
  title        = {Pith review of: Assessing AI-generated music detection in real-world broadcast monitoring},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LXIWS4V3}},
  note         = {Machine review of arXiv:2608.07359}
}
read the original abstract

The proliferation of AI-generated music in broadcast media raises concerns about transparency and fair compensation, but reliable detection under real broadcast conditions remains unresolved. Existing studies report substantial performance degradation in this domain, yet their evaluations are limited to synthetic broadcast data. To address this gap, we introduce BAMM (Broadcast AI-Music Monitoring), a 40-hour dataset of real-world television recordings containing AI-generated and human-made music. We compare clean-trained and broadcast-trained CNN variants across three progressively more challenging scenarios: Clean Foreground Music (CFM), Synthetic TV Broadcast (STB), and Real TV Broadcast (RTB). Both models achieve near-perfect performance on CFM but degrade substantially under synthetic broadcast conditions. Broadcast-oriented training improves robustness compared with clean training, although performance remains limited. On RTB, evaluated using BAMM, both models degrade further and show substantial score overlap between AI-generated and human-made music. These results expose a critical domain gap and show that current training approaches on CNN-based detectors remain insufficient for reliable AI-generated music detection in broadcast monitoring.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 29 canonical work pages

  1. [1]

    Assessing AI-generated music detection in real-world broadcast monitoring

    INTRODUCTION Generative music models have rapidly lowered the bar- rier to music production. Their widespread adoption has brought to light a range of technical and legal challenges concerning training data provenance, training data replica- tion, the definition of authorship, and compensation frame- works. These tensions have resulted in lawsuits against...

    work page Pith review arXiv 2026

  2. [2]

    Music-specific work has primarily focused on unintentionally left artifacts by generative systems

    RELA TED WORK The detection of AI-generated music has emerged within the MIR field as part of a broader effort to identify syn- thetic audio, alongside related tasks such as voice spoofing and synthetic speech detection [8, 9]. Music-specific work has primarily focused on unintentionally left artifacts by generative systems. Afchar et al. [6] showed that ...

    work page

  3. [3]

    BAMM DA TASET BAMM (Broadcast AI-Music Monitoring) is a novel dataset comprising 40 hours of real-world broadcast con- tent curated for AI-generated music detection. The dataset was constructed by identifying clean AI-generated and human-made reference tracks within a global broadcast archive using audio fingerprinting, followed by a multi- stage filterin...

    work page 2025

  4. [4]

    We compare two vari- ants of the architecture proposed by Afchar et al

    EXPERIMENTAL SETUP This benchmark examines how the training domain affects CNN-based detectors designed to identify AI-generated music in real broadcast settings. We compare two vari- ants of the architecture proposed by Afchar et al. [6], both operating on 5-second windows but trained under different conditions: clean foreground music and broadcast-orien...

    work page

  5. [5]

    The comparison betweenCNN CleanandCNN Broadcastallows us to analyze how the training domain influences the robustness of AI-generated music detection

    RESULTS We evaluate both models across three scenarios with in- creasing levels of broadcast complexity: Clean Fore- ground Music (CFM), Synthetic TV Broadcast (STB), and Real TV Broadcast (RTB). The comparison betweenCNN CleanandCNN Broadcastallows us to analyze how the training domain influences the robustness of AI-generated music detection. While the ...

    work page

  6. [6]

    CONCLUSIONS In this paper, we introduced BAMM, a dataset for AI- generated music detection designed to address key limi- tations of existing benchmarks. Unlike datasets focused primarily on clean or isolated musical excerpts, BAMM targets the real broadcast setting, where music appears in short segments, alternates between foreground and back- ground role...

    work page 2000

  7. [7]

    Cátedra IA y Música

    ACKNOWLEDGMENTS This work is supported by the "Cátedra IA y Música" project (TSI-100929-2023-1), funded by the Sec- retaría de Estado de Digitalización e Inteligencia Artificial, the European Union-Next Generation EU funds and BMAT Music Innovators. And by the "IMPA" project (PID2023-152250OB-I00) funded by MCIU/AEI/10.13039/501100011033/FEDER, UE

    doi:10.13039/501100011033/feder 2023

  8. [8]

    From Audio Deepfake Detection to AI-Generated Music De- tection – a pathway and overview,

    Y . Li, M. Milling, L. Specia, and B. W. Schuller, “From Audio Deepfake Detection to AI-Generated Music De- tection – a pathway and overview,”arXiv preprint arXiv:2412.00571, 2024

    arXiv 2024

Show all 31 references
  1. [9]

    Not All Deep- fakes Are Created Equal: Triaging Audio Forgeries for Robust Deepfake Singer Identification,

    D. Salvi, H. V . Koops, and E. Quinton, “Not All Deep- fakes Are Created Equal: Triaging Audio Forgeries for Robust Deepfake Singer Identification,”arXiv preprint arXiv:2510.17474, 2025

    2025

  2. [10]

    Major record companies sue Suno, Udio for ‘mass infringement’ of copyright,

    D. Tencer, “Major record companies sue Suno, Udio for ‘mass infringement’ of copyright,” https://www.musicbusinessworldwide.com/major- record-companies-sue-ai-music-generators-suno-udio- for-mass-infringement-of-copyright/, Jun. 2024

    2024

  3. [11]

    Warner Music Group strikes ‘landmark’ deal with Suno; settles copy- right lawsuit against AI music generator,

    M. Stassen, “Warner Music Group strikes ‘landmark’ deal with Suno; settles copy- right lawsuit against AI music generator,” https://www.musicbusinessworldwide.com/warner- music-group-settles-with-suno-strikes-first-of-its- kind-deal-with-ai-song-generator/, Nov. 2025

    2025

  4. [12]

    Technical Solutions for Marking and Detecting AI-generated Audio in the Context of Article 50 of the AI Act,

    X. Serra, R. O. Araz, R. Batlle-Roca, L. Juvela, D. López, and M. Rocamora, “Technical Solutions for Marking and Detecting AI-generated Audio in the Context of Article 50 of the AI Act,” Euro- pean Commission, Technical Report, 2026, tender EC- CNECT/2024/VLVP/0115 Watermarking Audio

    2026

  5. [13]

    Ai music tops 50% of daily uploads on deezer,

    J. Wendel, “Ai music tops 50% of daily uploads on deezer,” https://newsroom-deezer.com/2026/07/ ai-music-exceeds-50-percent-daily-uploads-deezer/, 2026, [Online; accessed 24-July-2026]

    2026

  6. [14]

    First, video killed the radio star, and now AI is going after video: A study on detect- ing GenAI music in broadcast audio,

    D. Bolboac ˘a, “First, video killed the radio star, and now AI is going after video: A study on detect- ing GenAI music in broadcast audio,” https://www. bmat.com/genai-music-broadcasting/, 2026, [Online; accessed 24-July-2026]

    2026

  7. [15]

    AI-generated music detection and its challenges,

    D. Afchar, G. Meseguer-Brocal, and R. Hennequin, “AI-generated music detection and its challenges,” in Proc. of the IEEE International Conference on Acous- tics, Speech and Signal Processing (ICASSP), 2025

    2025

  8. [16]

    AI-generated music detection in broadcast monitoring,

    D. López-Ayala, A. Cabello, P. Zinemanas, E. Molina, and M. Rocamora, “AI-generated music detection in broadcast monitoring,” inProc. of the 2026 IEEE Inter- national Conference on Acoustics, Speech and Signal Processing (ICASSP), 2026

    2026

  9. [17]

    Open broadcast media audio from tv: A dataset of tv broadcast audio with relative music loudness anno- tations,

    B. Meléndez-Catalán, E. Molina, and E. Gómez, “Open broadcast media audio from tv: A dataset of tv broadcast audio with relative music loudness anno- tations,”Transactions of the International Society for Music Information Retrieval (TISMIR), vol. 2, no. 1, pp. 43–51, 2019

    2019

  10. [18]

    Enhanced television broadcast monitor- ing with source separation-assisted audio fingerprint- ing: A case study,

    G. Cortès-Sebastià, M. Miron, E. Molina, A. Ciurana, and X. Serra, “Enhanced television broadcast monitor- ing with source separation-assisted audio fingerprint- ing: A case study,”Multimedia Tools and Applications, vol. 84, no. 42, pp. 50 595–50 628, 2025

    2025

  11. [19]

    High fidelity neural audio compression,

    A. Défossez, J. Copet, G. Synnaeve, and Y . Adi, “High fidelity neural audio compression,”Transactions on Machine Learning Research (TMLR), 2023

    2023

  12. [20]

    High-fidelity audio compression with im- proved RVQGAN,

    R. Kumar, P. Seetharaman, A. Luebs, I. Kumar, and K. Kumar, “High-fidelity audio compression with im- proved RVQGAN,” inProc. of the Conference on Neu- ral Information Processing Systems (NeurIPS), 2023

    2023

  13. [21]

    A Fourier Explanation of AI-music Artifacts,

    D. Afchar, G. Meseguer-Brocal, K. Akesbi, and R. Hennequin, “A Fourier Explanation of AI-music Artifacts,” inProceedings of the International Society for Music Information Retrieval Conference (ISMIR), 2025

    2025

  14. [22]

    Sonics: Synthetic or not - identify- ing counterfeit songs,

    M. A. Rahman, Z. I. A. Hakim, N. H. Sarker, B. Paul, and S. A. Fattah, “Sonics: Synthetic or not - identify- ing counterfeit songs,” inInternational Conference on Learning Representations (ICLR), 2025

    2025

  15. [23]

    Genius song lyrics,

    C. GDCJ, “Genius song lyrics,” https: //www.kaggle.com/datasets/carlosgdcj/ genius-song-lyrics-with-language-information, 2023, [Online; accessed 16-March-2026]

    2023

  16. [24]

    The AI Music Arms Race: On the Detec- tion of AI-Generated Music,

    L. C. Vila, B. L. T. Sturm, L. Casini, and D. Dal- mazzo, “The AI Music Arms Race: On the Detec- tion of AI-Generated Music,”Transactions of the In- ternational Society for Music Information Retrieval (TISMIR), vol. 8, no. 1, pp. 179–194, 2025

    2025

  17. [25]

    Relative music loudness esti- mation in TV broadcast audio using deep learning: An industrial perspective,

    B. Meléndez Catalán, “Relative music loudness esti- mation in TV broadcast audio using deep learning: An industrial perspective,” Ph.D. dissertation, Universitat Pompeu Fabra, Apr. 2021

    2021

  18. [28]

    BAF: An audio fingerprint- ing dataset for broadcast monitoring,

    G. Cortès, A. Ciurana, E. Molina, M. Miron, O. Mey- ers, J. Six, and X. Serra, “BAF: An audio fingerprint- ing dataset for broadcast monitoring,” inProc. of the International Society for Music Information Retrieval Conference (ISMIR), 2022

    2022

  19. [29]

    Da-TACOS: A dataset for cover song identification and understanding,

    F. Yesiler, C. Tralie, A. Correya, D. F. Silva, P. Tovsto- gan, E. Gómez, and X. Serra, “Da-TACOS: A dataset for cover song identification and understanding,” in Proc. of the International Society for Music Informa- tion Retrieval Conference (ISMIR), 2019

    2019

  20. [30]

    Music identification with audio fingerprinting an industrial perspective,

    G. Cortès Sebastià, “Music identification with audio fingerprinting an industrial perspective,” Ph.D. disser- tation, Universitat Pompeu Fabra, Feb. 2025

    2025

  21. [31]

    Free English and Czech telephone speech cor- pus shared under the CC-BY-SA 3.0 license,

    M. Korvas, O. Plátek, O. Dušek, L. Žilka, and F. Ju- rˇcíˇcek, “Free English and Czech telephone speech cor- pus shared under the CC-BY-SA 3.0 license,” inPro- ceedings of the International Conference on Language Resources and Evaluation (LREC), 2014

    2014

  22. [1024]

    The Mel- spectrogram is then segmented into discrete time windows to form the input tensor of the CNN

    The resulting power spectrogram is projected onto 128 Mel bands spanning 20 Hz to 4 kHz, converted to the logarithmic dB scale, and standardized with a fixed global mean of−4.0and standard deviation of3.0. The Mel- spectrogram is then segmented into discrete time windows to fo...

  23. [2026]

    proach introduced by Afchar et al

    Score values remain low before the release of Suno v3.5, and increase afterward, indicating the appearance of Suno-generated music. proach introduced by Afchar et al. [12]. It operates on artifact-fingerprint features extracted from audio resam- pled to 16 kHz. The fifth model...

    2024

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.