AI Music Library Français
Exclusive sale · sold once

100,000+ AI music tracks.
One owner.

Produced with Suno v5.5, under the terms in force before 3 September 2026: the ones that assigned to the subscriber every right in their generations. The catalog sells once, then it comes off the market.

Request the brief → Listen to excerpts
Answer within 48 hours. No commitment.
100,000+tracks
5,000+ hof music
35dimensions per track
1,280vector dimensions

344 moods, 2,541 instruments, 19 commercial settings

Every track is described across 35 dimensions.

A catalog delivered without description is a pile of files: you have to listen to all of it before you can do anything with it. This one arrives described, at a depth you will not find in a subscription library.

27

measurements taken from the signal

Values extracted from the audio by Essentia, not labels someone declared.

tempo bpm Beats per minute, estimated from the signal rather than declared. tempo confidence bpm_confidence Reliability of the tempo estimate. A low value flags a rubato track or one with no clear pulse. key root key_root Root note of the detected key, from C to B. key scale key_scale Major or minor. Combined with the root, it gives the full key. key strength key_strength How clearly the tonal centre asserts itself, 0 to 1. Atonal or heavily modulating tracks score low. LUFS loudness loudness_lufs Perceived loudness under EBU R128, in integrated LUFS. This is what lets you level a whole catalog without listening to it. loudness range loudness_range Spread between quiet and loud passages. A wide range sits badly as background music in a shop. energy energy Perceived intensity from 0 to 1, dominated by loudness. danceability (signal) danceability_basic Rhythmic regularity computed directly from the waveform. danceability (model) danceability_ml Danceability predicted by the Discogs-Effnet network, trained on human annotations. top genre genre_top Most probable genre among the 400 labels of the Discogs model. genre confidence genre_confidence Probability attached to the top genre. Below 0.3, the label is debatable. candidate genres genre_top5 The genre hypotheses kept above the confidence threshold, ranked, up to five. A clear-cut track keeps one, a borderline track keeps five: the count is itself a diagnosis. mood happy mood_happy Probability of a happy character, from a dedicated classifier. mood party mood_party Probability of a party character. This is what separates aperitivo from dinner service. mood relaxed mood_relaxed Probability of a relaxed character, the backbone of spa and reception settings. valence valence Overall positivity derived from the moods: 0.6 x happy + 0.4 x relaxed. vocal or instrumental voice_instrumental Whether the track carries singing. Decisive for a shop that wants no lyrics during advice hours. vocal confidence voice_instrumental_confidence Reliability of that detection. Choirs treated as a pad bring it down. raw duration duration_sec Total file duration, silence included. head silence silence_head_sec Blank before the first sound. This is what creates gaps between tracks when nobody measures it. tail silence silence_tail_sec Blank after the last sound, essential for timing a tight crossfade. effective duration effective_duration_sec Actually sounding duration, silence deducted. This is the one that matters when filling a slot. tonal stability tonal_stability 0 if the track modulates a lot, 1 if it stays on one centre. Useful for segueing without a harmonic clash. loudness peaks loudness_peaks_count Number of marked surges in the track. peak density loudness_peaks_density Peaks per minute. A retail floor wants few, a gym wants many. peak timestamps loudness_peaks_timestamps Exact second of every peak. This is what lets you drop an announcement between surges rather than across one.
8

editorial fields on top

Where measurement stops, description takes over. The vocabulary is CLOSED and cut for retail: 24 families, 19 trading moments, 5 vocal types. Every track is described by crossing two independent sources, the production intent and the acoustic measurements; when they disagree, the audio settles energy, the intent settles genre, and the conflict is flagged — that happens on 15% of tracks. Any value outside the list is refused on write, so no invented category ever enters the catalog. A second pass then re-reads every set of tags, not to rewrite it but to catch it out: across the reviews run, 2% needed correcting.

primary genre24 families344 moods2,541 instruments19 commercial settingsenergy bandvocal typeconfidence score
1,280

acoustic vector dimensions

The full fingerprint of every track. Enough to ask for "tracks that sound like this one" without writing a single rule, on delivery day rather than six months later.

similarity searchautomatic playlistsduplicate detectionrecommendationgrouping by ear

The catalog ships described, in this format, track by track. The vocabularies are closed: 24 families, 19 commercial settings, 5 vocal types.

Listen.

Six excerpts from the catalog, with the real tags exactly as they ship. Forty seconds each.

My Valentine
Jazz/Swing
romanticvelvetyelegant pianoupright bass 122 BPM female 3 min 25
See every piece of metadata for this track
27 measurements taken from the signalThese measurements cover the whole track and ship alongside every file. The excerpt above plays only forty seconds of it: a modulating passage can therefore sound in a different key from the piece itself.
tempobpm
122.4
tempo confidencebpm_confidence
0.359
key rootkey_root
C
key scalekey_scale
major
key strengthkey_strength
0.883
LUFS loudnessloudness_lufs
-15.14
loudness rangeloudness_range
6.91
energyenergy
0.32
danceability (signal)danceability_basic
0.333
danceability (model)danceability_ml
0.182
top genregenre_top
Jazz---Cool Jazz
genre confidencegenre_confidence
0.203
candidate genresgenre_top5
Cool Jazz 20%Smooth Jazz 18%
mood happymood_happy
0.955
mood partymood_party
0.016
mood relaxedmood_relaxed
0.988
valencevalence
0.396
vocal or instrumentalvoice_instrumental
voice
vocal confidencevoice_instrumental_confidence
0.75
raw durationduration_sec
205.12
head silencesilence_head_sec
0.1
tail silencesilence_tail_sec
0.5
effective durationeffective_duration_sec
204.52
tonal stabilitytonal_stability
0.4
loudness peaksloudness_peaks_count
1
peak densityloudness_peaks_density
0.005
peak timestampsloudness_peaks_timestamps
28.0 s
8 editorial fields
primary genregenre_primary
Jazz/Swing
secondary genresgenre_secondary
Seasonal/Christmas
moodsmoods
romanticvelvetyelegantnostalgic
commercial settingsambiances
dinnerevening loungechristmasfine dining
energy bandenergy_band
low
vocal typevocal
female
instrumentationinstrumentation
pianoupright bassbrushes
confidence scoreconfidence
0.85
Acoustic vector
1,280dimensions
[-0.0230, 0.0006, -0.0154, 0.0028, 0.0781, 0.2313, 0.0267, 0.0258, 0.0439, -0.0014, 0.1567, 0.2304, …]

First twelve values out of 1,280. The full vector ships with the catalog: it is what makes similarity search work without writing a rule.

Gets lighter together
Soul/R&B
warmgentlevelvety pianobass 101 BPM male 3 min 33
See every piece of metadata for this track
27 measurements taken from the signalThese measurements cover the whole track and ship alongside every file. The excerpt above plays only forty seconds of it: a modulating passage can therefore sound in a different key from the piece itself.
tempobpm
101.3
tempo confidencebpm_confidence
1.407
key rootkey_root
G
key scalekey_scale
major
key strengthkey_strength
0.857
LUFS loudnessloudness_lufs
-13.12
loudness rangeloudness_range
5.58
energyenergy
0.487
danceability (signal)danceability_basic
0.37
danceability (model)danceability_ml
0.869
top genregenre_top
Funk / Soul---Contemporary R&B
genre confidencegenre_confidence
0.286
candidate genresgenre_top5
Contemporary R&B 29%RnB/Swing 28%Neo Soul 21%Soul 17%
mood happymood_happy
0.771
mood partymood_party
0.221
mood relaxedmood_relaxed
0.821
valencevalence
0.478
vocal or instrumentalvoice_instrumental
voice
vocal confidencevoice_instrumental_confidence
0.937
raw durationduration_sec
213.2
head silencesilence_head_sec
0.1
tail silencesilence_tail_sec
1.2
effective durationeffective_duration_sec
211.9
tonal stabilitytonal_stability
0.8
loudness peaksloudness_peaks_count
0
peak densityloudness_peaks_density
0
peak timestampsloudness_peaks_timestamps
—
8 editorial fields
primary genregenre_primary
Soul/R&B
secondary genresgenre_secondary
Pop
moodsmoods
warmgentlevelvetysoothing
commercial settingsambiances
morning caféevening loungereception & waitingdinner
energy bandenergy_band
medium
vocal typevocal
male
instrumentationinstrumentation
pianobasssoft percussionstring pad
confidence scoreconfidence
0.78
Acoustic vector
1,280dimensions
[-0.0140, 0.0199, 0.0102, 0.0186, -0.0469, 0.0043, 0.1632, -0.0221, 0.0297, -0.0052, 0.5202, 0.0054, …]

First twelve values out of 1,280. The full vector ships with the catalog: it is what makes similarity search work without writing a rule.

Nuit de Figuier
Deep/Chill House
warmsoothinggroovy brasskeys 118 BPM instrumental 3 min 38
See every piece of metadata for this track
27 measurements taken from the signalThese measurements cover the whole track and ship alongside every file. The excerpt above plays only forty seconds of it: a modulating passage can therefore sound in a different key from the piece itself.
tempobpm
117.7
tempo confidencebpm_confidence
3.214
key rootkey_root
A
key scalekey_scale
minor
key strengthkey_strength
0.916
LUFS loudnessloudness_lufs
-12.83
loudness rangeloudness_range
5.93
energyenergy
0.507
danceability (signal)danceability_basic
0.399
danceability (model)danceability_ml
0.985
top genregenre_top
Electronic---Deep House
genre confidencegenre_confidence
0.432
candidate genresgenre_top5
Deep House 43%House 26%Nu-Disco 22%Tech House 17%Downtempo 13%
mood happymood_happy
0.913
mood partymood_party
0.341
mood relaxedmood_relaxed
0.774
valencevalence
0.561
vocal or instrumentalvoice_instrumental
instrumental
vocal confidencevoice_instrumental_confidence
0.94
raw durationduration_sec
219.6
head silencesilence_head_sec
0
tail silencesilence_tail_sec
1.8
effective durationeffective_duration_sec
217.8
tonal stabilitytonal_stability
1
loudness peaksloudness_peaks_count
0
peak densityloudness_peaks_density
0
peak timestampsloudness_peaks_timestamps
—
8 editorial fields
primary genregenre_primary
Deep/Chill House
secondary genresgenre_secondary
—
moodsmoods
warmsoothinggroovydreamy
commercial settingsambiances
summer terraceevening loungeaperitivobrunch
energy bandenergy_band
medium
vocal typevocal
instrumental
instrumentationinstrumentation
brasskeyssubpercussion
confidence scoreconfidence
0.88
Acoustic vector
1,280dimensions
[-0.0062, 0.1089, -0.0282, 0.0949, 0.0763, -0.0263, 0.3602, -0.0185, 0.0415, 0.0030, 0.3576, 0.3696, …]

First twelve values out of 1,280. The full vector ships with the catalog: it is what makes similarity search work without writing a rule.

Souffle Voluptueux
Electro Swing
festiveplayfulgroovy finger-snap percussionwalking bass 107 BPM male 3 min 03
See every piece of metadata for this track
27 measurements taken from the signalThese measurements cover the whole track and ship alongside every file. The excerpt above plays only forty seconds of it: a modulating passage can therefore sound in a different key from the piece itself.
tempobpm
107
tempo confidencebpm_confidence
2.679
key rootkey_root
A
key scalekey_scale
minor
key strengthkey_strength
0.956
LUFS loudnessloudness_lufs
-14.26
loudness rangeloudness_range
4.53
energyenergy
0.65
danceability (signal)danceability_basic
0.41
danceability (model)danceability_ml
0.978
top genregenre_top
Electronic---Tropical House
genre confidencegenre_confidence
0.262
candidate genresgenre_top5
Tropical House 26%Europop 22%Disco Polo 21%Dance-pop 17%Euro House 15%
mood happymood_happy
0.322
mood partymood_party
0.866
mood relaxedmood_relaxed
0.414
valencevalence
0.61
vocal or instrumentalvoice_instrumental
voice
vocal confidencevoice_instrumental_confidence
0.915
raw durationduration_sec
184.88
head silencesilence_head_sec
0.1
tail silencesilence_tail_sec
1.7
effective durationeffective_duration_sec
183.08
tonal stabilitytonal_stability
1
loudness peaksloudness_peaks_count
0
peak densityloudness_peaks_density
0
peak timestampsloudness_peaks_timestamps
—
8 editorial fields
primary genregenre_primary
Electro Swing
secondary genresgenre_secondary
—
moodsmoods
festiveplayfulgroovyupbeat
commercial settingsambiances
aperitivofestive eventbrunchfashion retail
energy bandenergy_band
high
vocal typevocal
male
instrumentationinstrumentation
finger-snap percussionwalking bassrockabilly guitarpiano
confidence scoreconfidence
0.76
Acoustic vector
1,280dimensions
[-0.0174, 0.0185, -0.0238, -0.0219, 0.0718, 0.0097, 0.3066, -0.0007, 0.0042, 0.0044, 0.4979, 0.0062, …]

First twelve values out of 1,280. The full vector ships with the catalog: it is what makes similarity search work without writing a rule.

Nocturne in Sage
Classique/Piano
soothingbrighttender violinstring ensemble 128 BPM instrumental 2 min 34
See every piece of metadata for this track
27 measurements taken from the signalThese measurements cover the whole track and ship alongside every file. The excerpt above plays only forty seconds of it: a modulating passage can therefore sound in a different key from the piece itself.
tempobpm
127.8
tempo confidencebpm_confidence
1.296
key rootkey_root
Bb
key scalekey_scale
major
key strengthkey_strength
0.925
LUFS loudnessloudness_lufs
-13.93
loudness rangeloudness_range
10.01
energyenergy
0.282
danceability (signal)danceability_basic
0.266
danceability (model)danceability_ml
0.087
top genregenre_top
Electronic---New Age
genre confidencegenre_confidence
0.226
candidate genresgenre_top5
New Age 23%Ambient 18%Celtic 17%Folk 14%Soundtrack 6%
mood happymood_happy
0.987
mood partymood_party
0.007
mood relaxedmood_relaxed
0.994
valencevalence
0.417
vocal or instrumentalvoice_instrumental
instrumental
vocal confidencevoice_instrumental_confidence
0.766
raw durationduration_sec
154.8
head silencesilence_head_sec
0.8
tail silencesilence_tail_sec
0.5
effective durationeffective_duration_sec
153.5
tonal stabilitytonal_stability
1
loudness peaksloudness_peaks_count
0
peak densityloudness_peaks_density
0
peak timestampsloudness_peaks_timestamps
—
8 editorial fields
primary genregenre_primary
Classical/Piano
secondary genresgenre_secondary
Ambient/Cinematic
moodsmoods
soothingbrighttenderintimate
commercial settingsambiances
spa & wellnessreception & waitinghome & decor retailevening lounge
energy bandenergy_band
low
vocal typevocal
instrumental
instrumentationinstrumentation
violinstring ensembleviolacello
confidence scoreconfidence
0.72
Acoustic vector
1,280dimensions
[-0.0335, -0.0173, 0.0289, -0.0103, 0.1305, 0.1883, 0.0419, 0.2301, -0.0037, -0.0021, 0.0115, 0.1303, …]

First twelve values out of 1,280. The full vector ships with the catalog: it is what makes similarity search work without writing a rule.

Muted Glow
Tropical House
brightdreamygentle Balearic flutefiltered pads 124 BPM instrumental 3 min 27
See every piece of metadata for this track
27 measurements taken from the signalThese measurements cover the whole track and ship alongside every file. The excerpt above plays only forty seconds of it: a modulating passage can therefore sound in a different key from the piece itself.
tempobpm
124
tempo confidencebpm_confidence
2.97
key rootkey_root
G
key scalekey_scale
minor
key strengthkey_strength
0.865
LUFS loudnessloudness_lufs
-13.58
loudness rangeloudness_range
5.81
energyenergy
0.476
danceability (signal)danceability_basic
0.419
danceability (model)danceability_ml
0.996
top genregenre_top
Electronic---Progressive House
genre confidencegenre_confidence
0.481
candidate genresgenre_top5
Progressive House 48%Progressive Breaks 29%Deep House 25%Deep Techno 19%Tech House 14%
mood happymood_happy
0.939
mood partymood_party
0.597
mood relaxedmood_relaxed
0.744
valencevalence
0.59
vocal or instrumentalvoice_instrumental
instrumental
vocal confidencevoice_instrumental_confidence
0.897
raw durationduration_sec
208.92
head silencesilence_head_sec
0.6
tail silencesilence_tail_sec
1.4
effective durationeffective_duration_sec
206.92
tonal stabilitytonal_stability
0.4
loudness peaksloudness_peaks_count
0
peak densityloudness_peaks_density
0
peak timestampsloudness_peaks_timestamps
—
8 editorial fields
primary genregenre_primary
Tropical House
secondary genresgenre_secondary
Deep/Chill House
moodsmoods
brightdreamygentlegroovy
commercial settingsambiances
summer terraceevening loungefashion retailaperitivo
energy bandenergy_band
medium
vocal typevocal
instrumental
instrumentationinstrumentation
Balearic flutefiltered padsoffbeat clapssub bass
confidence scoreconfidence
0.72
Acoustic vector
1,280dimensions
[-0.0126, 0.1276, -0.0445, 0.0598, 0.0250, -0.0214, 0.4396, 0.0146, 0.0223, -0.0010, 0.2426, 0.2712, …]

First twelve values out of 1,280. The full vector ships with the catalog: it is what makes similarity search work without writing a rule.

40-second excerpts. The catalog ships as complete files.

What you buy is genuinely yours.

The question a buyer asks last is the one that decides everything: what exactly does the right to sell this catalog rest on. Here is the answer, in three pieces you can verify before signing.

The clause that assigns the rights

The generation service's terms, in their version predating 3 September 2026, assigned to paid-tier subscribers every right in their outputs. Every generation in this catalog sits on that tier, invoice by invoice. Those terms are no longer online: they are archived, timestamped and hashed, and they are the ones that govern the catalog.

A unique identifier inside every file

Every track carries, in its own metadata, the identifier assigned at generation. This is not a separate ledger you have to take on trust: open a file at random and it is in there. Measured on 400 randomly drawn files, 99.5% carry one and none is duplicated.

No licence to take over

A work produced entirely by a machine, with no human creative input, carries no copyright under French law, for want of an author. Exploiting this catalog therefore requires no licence from the generation service and no subscription to maintain after purchase. You inherit no dependency.

What the catalog contains

A finished asset, not a subscription and not a service. MP3 files at 193 kbps for roughly 440 GB, metadata written into every file, and a provenance record carrying, track by track, the prompt that made it, its date, its model and its tempo. No fragments and no loops: the generator discarded short clips at production time, average length runs 3 min 02 s and the shortest track in the catalog runs 1 min 12 s. And tracks end, they are not cut off: measured on a random draw, 87% fade out at least twelve decibels below their own average level, with a median gap of thirty-three decibels. A clean ending is what separates a library you can broadcast from a pile of files you still have to edit.

Take the lead on volume

This catalog is for anyone who wants to move ahead on quantity without giving up quality, and without starting from scratch. A hundred thousand tracks arrive ready to drop in: described across thirty-five dimensions, sorted by commercial setting, tagged properly at production time rather than patched afterwards. You plug it in, you filter, you build custom playlists the same day, and you layer your own tags on top without having to undo ours. In-store music operators, app and streaming publishers, agencies, video platforms, game studios, retail media networks: the six months of indexing work is already done.

Buying the catalog

The sale is exclusive. What leaves is not resold, not kept in catalog and not exploited elsewhere. It covers the whole library or coherent batches, organised by mood or by use, and every batch sold is withdrawn from sale. Price depends on scope: tell us the intended use and the volume, and we will propose both the split and the figure.

Frequently asked questions

Is this music royalty-free?

The answer is more precise than "royalty-free". The rights in these outputs were assigned to the producer by the generation service, under the terms in force on the date of production. The producer assigns them to you in turn, exclusively. You are not buying a usage licence, you are buying the asset.

Can we resell or sublicense the tracks?

Yes. The sale transfers the exploitation rights, which includes sublicensing and resale. That is the substantive difference from a subscription library, where use is granted but never transferred. The exact scope sits in the assignment agreement, supplied with the brief.

Can we use it in our own app or in-store music service?

That is the main intended use. The catalog ships as files with their metadata, ready to feed an in-store music service, an application or a platform, with no recurring royalty and no play counter.

How much does an AI music catalog cost?

No price is shown because it depends on scope: the whole library or batches, full exclusivity or exclusivity per vertical. Describe the intended use in the form and we answer within 48 hours with a figure and what it covers.

Do these tracks have to be declared to a collecting society?

A work produced entirely by a machine, with no human creative input, carries no copyright under French law, for want of an author. That is what the provenance record is there to establish. The legal note in the brief sets out the reasoning and its limits.

Can we listen before buying?

Yes. A representative sample is provided on request, together with its extract of the provenance record, so you can check the chain end to end before committing to anything.

Request the brief

Tell us the intended use and the volume you have in mind. We answer with the catalog breakdown, an audio sample and the legal note.