Which AI music detectors exist? The five cards that state music among the media they score, the generators each names, and the zero study records behind them
Five of the 72 AI detectors in the index state music among the media they score, by the modalities field of every card read 10 November 2026 in the data file: Ircam Amplify's AIMD, which detects nothing else and lists Suno versions 3 to 6, Udio, Lyria, Mureka, Riffusion, Treblo, ElevenLabs Music and ACE-Step on its model coverage page (docs.ircamamplify.io, read 10 November 2026); Hive, whose music attribution head returns 14 source classes from suno and udio to ace_step (docs.thehive.ai, read 8 November 2026); Resemble Detect, which lists Suno and Udio among its covered audio generators (resemble.ai, read 9 November 2026); Sightengine, which sells AI speech and music detection on its Pro plan (sightengine.com, read 9 November 2026); and AI or Not, whose Pro plan unlocks music detection (aiornot.com, read 9 November 2026). None of the 35 study records in the index scored any of the five on music.
Music is the youngest modality in the index and the least tested. A card enters the table below when its modalities field, taken from the vendor's own product page or documentation, names music or AI-generated tracks as something the detector scores; a speech or voice detector that would also receive a song is not counted, because the vendor does not say it scores one. Rows name the generators the vendor lists for music, whether a price is published, and how many study records in the index tested the card on music, which is zero in every row. This page is one of the questions answered from the index; the content type view lists every card by every modality it states, the studies by content type ranking counts the records per modality, and the AI detector register carries every card.
| Detector | Music as the card states it | Music generators the vendor names | Pricing published | Study records on music | Source | Read date |
|---|---|---|---|---|---|---|
| Ircam Amplify | "music (AI-generated tracks)", the product's only modality | Suno versions 3 to 6, Udio, Lyria, Mureka, Riffusion, Treblo (formerly Sonauto), ElevenLabs Music versions 1 and 2, ACE-Step; Boomy, LDM2 and MusicGen as opt-in legacy generators (AIMD V2.3) | No: access by application, a pay-as-you-go plan "from 2,000 tracks" and an Enterprise contract, with no price per track published | 0 | docs.ircamamplify.io | 10 November 2026 |
| Hive | "music" beside text, image, video and voice | Music attribution classes: duobao, google, heartmula, loudly, minimax, mubert, mureka, musicgen, riffusion, stable_audio_open, suno, udio, yue, ace_step (14) | No: "AI Music Detection" is Contact Sales on the pricing page | 0 (2 records, both on images) | docs.thehive.ai | 8 November 2026 |
| Resemble Detect | "music (Suno and Udio listed as covered)" beside voice, image and video | Suno, Udio, inside an audio list of 22 named generators "and over 100 more" | Yes: Flex at $0 a month with audio at $0.035 a second; Team $350 a month with audio at $0.015 a second | 0 | resemble.ai | 9 November 2026 |
| Sightengine | "music" beside image, video, deepfake and voice (audio speech) | Not named for music | Yes: "AI speech and music detection" on Pro at $99 a month for 40,000 operations | 0 (1 record, on images) | sightengine.com | 9 November 2026 |
| AI or Not | "music" beside image, video, voice and text | Not named for music | Yes: Pro at $5 a month "unlocks text, image, deepfake, voice, music and video detection"; $0.005 per second of audio pay as you go | 0 (1 record, on images) | aiornot.com | 9 November 2026 |
Modalities, generators and plans as each card records them from the page linked, read on the date shown; the index has not submitted a track to any of the five.
Which AI music detectors name the generators they cover?
Three of the five, and two name none. Ircam Amplify publishes the longest list and dates it: the AIMD V2.3 model coverage page, read 10 November 2026, marks ACE-Step, ElevenLabs Music versions 1 and 2 and Suno versions 3 to 6 as new and Lyria, Mureka, Riffusion, Treblo and Udio as existing, and the release note moves Boomy, LDM2 and MusicGen to an opt-in legacy flag because "several older generators have effectively disappeared from current production usage, while continuing to contribute disproportionately to false-positive detections" (docs.ircamamplify.io, read 10 November 2026). Hive's documentation lists 14 music attribution classes, from suno and udio to mubert, loudly and ace_step, returned by a source head beside the binary verdict (docs.thehive.ai, read 8 November 2026). Resemble Detect names Suno and Udio inside an audio generator list that is otherwise text-to-speech systems, from ElevenLabs to VALL-E 2 (resemble.ai, read 9 November 2026). Sightengine's generator list is for images and video, with Nano Banana, Midjourney and Sora among its entries, and its music detection is one line on the Pro plan (sightengine.com, read 9 November 2026); AI or Not's generator list covers images and text, and music appears only as a plan inclusion (aiornot.com, read 9 November 2026).
A named generator list is a coverage claim, not a result. Ircam Amplify's home page states 99% (accuracy) and under 1% false positives, and its V2.3 release note gives 99.7% (accuracy) with a 0.14% false positive rate at the 50% threshold on "internal benchmark datasets" whose size and composition are not published (ircamamplify.com, read 10 November 2026); Resemble Detect's "Up to 99.5% (accuracy)" is a pooled audio figure from the Podonos Audio DFD Bench, not a music figure (resemble.ai, read 9 November 2026); AI or Not's 98.9% (accuracy) is one figure for every modality (aiornot.com, read 9 November 2026); Hive and Sightengine publish no number. The claim vs evidence gap ranking cannot place any of the five on music because no study supplies the evidence side.
What does a music check cost, and what can you upload?
Three of the five publish a price and two do not. Resemble Detect's Flex plan has no subscription fee and bills $0.035 a second of audio, so a four-minute track costs $8.40 at that rate and $3.60 on the Team plan's $0.015 a second, which adds $350 a month (resemble.ai, read 9 November 2026). AI or Not bills $0.005 a second of audio on Pro, $1.20 for the same four minutes, inside a $5 a month plan that carries $10 of credits (aiornot.com, read 9 November 2026). Sightengine counts operations rather than seconds, 40,000 a month on the $99 Pro plan that carries the music model (sightengine.com, read 9 November 2026). Hive's AI Music Detection line is Contact Sales (deploy.com, read 8 November 2026), and Ircam Amplify reviews each application before a demo and sells tracks in pay-as-you-go blocks from 2,000 with no unit price on any page read (ircamamplify.com, read 10 November 2026). The free API tiers question notes that AI or Not's and Sightengine's free allowances cover their APIs, and the API pricing ranking prices text, not audio, so the per-second rates above stay on this page.
Audio inputs the five cards state
| Detector | Formats stated | Duration and size ceilings | Unit of scoring | Source | Read date |
|---|---|---|---|---|---|
| Ircam Amplify | AIF, AIFF, FLAC, M4A, MP3, MP4, OGG, WAV, WEBM (API guide) | 3 seconds to 30 minutes a track; 20 files a request; 256 kbps MP3 recommended | A 0 to 100 probability per track, a suspected generator above 50%, optional time analysis per segment | docs.ircamamplify.io | 10 November 2026 |
| Hive | flac, mp3, ogg, wav, m4a plus the video formats | File size and media length depend on project configuration | Music scored per 3-second chunk; speech per 10-second chunk | docs.thehive.ai | 8 November 2026 |
| Resemble Detect | wav, mp3, m4a, ogg, aac, flac, amr, 3gp, 3gpp plus the video formats | Direct upload up to 150 MB; a public URL with no size limit; a token for larger files | Windows of 1 to 4 seconds, a region selectable with start and end | docs.resemble.ai | 9 November 2026 |
| Sightengine | Audio formats not listed on the pages read | Audio 12 MB on Free, 20 MB on paid plans | An operation per call; a confidence from 0 to 1 | sightengine.com | 9 November 2026 |
| AI or Not | "common formats such as MP3 and WAV" (API); MP3 on the web tool | One file a check on the web tool; responses of up to two minutes on long files | A confidence score per file | docs.aiornot.com | 9 November 2026 |
Inputs as the input limits field of each card records them, read on the date shown. Ircam Amplify's 30-minute ceiling is the only stated duration limit among the five; the file types ranking counts Resemble Detect's 19 extensions and Ircam Amplify's 9 API guide formats from 10 November 2026 (17 and 7 until the aif, aiff, amr and 3gp tokens were counted under the stated rule, a correction logged on the corrections page).
Has any study tested an AI music detector?
No record in the index has. The audio records are speech challenges: ASVspoof 5 scored 53 anonymised teams on 542,086 spoofed utterances of read speech, and SAFE 2025 scored twelve teams on 2,600 clips from 13 text-to-speech models, and neither names a card in this index or contains a song (arXiv 2601.03944, arXiv 2510.03387, read 9 November 2026). The three cards with any record, Hive, Sightengine and AI or Not, were scored on images in Ha and Passananti 2024 and RAID image 2025, where AI or Not under its earlier name Optic recorded a 24.47% false positive rate on 280 human artworks (DOI 10.1145/3658644.3670306, Table 1, read 9 November 2026); an image result says nothing about a music model. Ircam Amplify and Resemble Detect have no record at all. A reader choosing among the five is therefore choosing among vendor statements, and the voice and video question shows the same gap for the nine cards that state both voice and video.
Every accuracy figure attached to an AI music detector in this index is the vendor's own, measured on a set the vendor assembled and did not publish, and no outside study has scored a single song.
Questions about AI music detectors
Is there a detector for AI-generated music?
5 of 72 cards state music among the media they score, read 10 November 2026: Ircam Amplify (music only), Hive, Resemble Detect, Sightengine and AI or Not (docs.ircamamplify.io, docs.thehive.ai, resemble.ai, sightengine.com, aiornot.com, read 8 to 10 November 2026).
Which detector covers Suno and Udio?
3 of the 5 name both: Ircam Amplify (Suno versions 3 to 6 and Udio), Hive (suno and udio among 14 attribution classes) and Resemble Detect (Suno and Udio in its audio list), by the generators field of each card read 10 November 2026; Sightengine and AI or Not name no music generator.
Which AI music detector is free to try?
$0 a month is the entry price on three cards, read 9 November 2026, with the music check metered or gated: Resemble Detect's Flex plan bills $0.035 a second of audio from the first credit (resemble.ai), AI or Not's Free plan gives $5 in credits but music detection is listed under the $5 a month Pro plan (aiornot.com), and Sightengine's Free Tier of 2,000 operations covers image and video detection while music detection sits on the $99 Pro plan (sightengine.com).
How accurate are AI music detectors?
No study in the index has measured one: 0 of the 35 records scored a card on music, read 10 November 2026. The vendor figures are Ircam Amplify's 99.7% (accuracy) and 0.14% false positive rate at the 50% threshold on internal datasets of unstated size (ircamamplify.com, read 10 November 2026) and Resemble Detect's pooled audio figure of up to 99.5% (accuracy) on the Podonos bench (resemble.ai, read 9 November 2026).
How long a track can an AI music detector take?
3 seconds to 30 minutes on Ircam Amplify, the only card that states a duration range, with 20 files a request (docs.ircamamplify.io, read 10 November 2026). Resemble Detect caps a direct upload at 150 MB and takes a public URL with no size limit (docs.resemble.ai, read 9 November 2026).
Sources
All read 10 November 2026 unless an earlier date is shown; the source of every row is linked in the tables.
- Ircam Amplify home page: https://www.ircamamplify.com/
- Ircam Amplify AIMD model coverage: https://docs.ircamamplify.io/api-docs/aimd-guide/model-coverage
- Ircam Amplify audio input methods: https://docs.ircamamplify.io/api-docs/aimd-guide/audio-input-methods
- Ircam Amplify request access: https://www.ircamamplify.com/request-aimd
- Hive AI-generated content detection docs: https://docs.thehive.ai/docs/ai-generated-content-detection (read 8 November 2026)
- Hive image, video and music attribution classes: https://docs.thehive.ai/docs/ai-image-and-video-detection (read 8 November 2026)
- Hive supported file formats: https://docs.thehive.ai/docs/supported-file-formats (read 8 November 2026)
- Deploy (Hive) pricing: https://deploy.com/pricing (read 8 November 2026)
- Resemble Detect product page: https://www.resemble.ai/products/detect (read 9 November 2026)
- Resemble AI pricing: https://www.resemble.ai/pricing (read 9 November 2026)
- Resemble AI benchmarks: https://www.resemble.ai/benchmarks (read 9 November 2026)
- Resemble Detect API, create a detection: https://docs.resemble.ai/detect/create (read 9 November 2026)
- Sightengine pricing: https://sightengine.com/pricing (read 9 November 2026)
- Sightengine AI image detector (generators): https://sightengine.com/detect-ai-generated-images (read 9 November 2026)
- AI or Not pricing: https://www.aiornot.com/pricing (read 9 November 2026)
- AI or Not home page (claim): https://www.aiornot.com/ (read 9 November 2026)
- AI or Not image detector (generators): https://www.aiornot.com/ai-image-detector (read 9 November 2026)
- AI or Not API reference (inputs): https://docs.aiornot.com/api-reference/reports-by-modality/image (read 9 November 2026)
- ASVspoof 5 evaluation (Wang et al. 2026): https://arxiv.org/abs/2601.03944 (read 9 November 2026)
- SAFE challenge (Trapeznikov et al. 2025): https://arxiv.org/abs/2510.03387 (read 9 November 2026)
- Ha and Passananti 2024: https://doi.org/10.1145/3658644.3670306 (read 9 November 2026)
Revision log: 2026-11-10: page created from the modalities, generators, plans and input limits fields of 72 cards at index version 2026.11.12.
Related pages: Which AI detectors are free for nonprofits or education? The 12 cards that sell an institutional programme, the one university platform that charges nothing, and the two nonprofit tools that closed, Is there a 100% accurate AI detector? What published tests show, AI detector accuracy claims vs published results, ranked by the gap and Questions about AI detectors, answered from the index.