Dataset Viewer
Auto-converted to Parquet Duplicate
audio
audio
film_id
string
component
string
kind
string
channels
int64
sample_rate
int64
bit_depth
int64
frames
int64
duration_sec
float64
peak
float64
mix_type
string
license
string
film_01
dialogue
target_stem
2
48,000
24
224,640,000
4,680
0.559804
additive
CC-BY-SA-4.0
film_01
music
target_stem
2
48,000
24
224,640,000
4,680
0.679827
additive
CC-BY-SA-4.0
film_01
sfx
target_stem
2
48,000
24
224,640,000
4,680
0.891251
additive
CC-BY-SA-4.0
film_01
linear_mix
input_mix_additive
2
48,000
24
224,640,000
4,680
0.89125
additive
CC-BY-SA-4.0
film_01
dx
fine_stem
2
48,000
24
224,640,000
4,680
0.353607
additive
CC-BY-SA-4.0
film_01
vo
fine_stem
2
48,000
24
224,640,000
4,680
0.559811
additive
CC-BY-SA-4.0
film_01
mx
fine_stem
2
48,000
24
224,640,000
4,680
0.679828
additive
CC-BY-SA-4.0
film_01
fx
fine_stem
2
48,000
24
224,640,000
4,680
0.879004
additive
CC-BY-SA-4.0
film_01
fol
fine_stem
2
48,000
24
224,640,000
4,680
0.333315
additive
CC-BY-SA-4.0
film_01
bg
fine_stem
2
48,000
24
224,640,000
4,680
0.075771
additive
CC-BY-SA-4.0
film_01
pfx
fine_stem
2
48,000
24
224,640,000
4,680
0.457258
additive
CC-BY-SA-4.0
film_01
arch
fine_stem
2
48,000
24
224,640,000
4,680
0.35969
additive
CC-BY-SA-4.0
film_02
dialogue
target_stem
2
48,000
24
34,864,000
726.333
0.247578
non_additive
CC-BY-4.0
film_02
music
target_stem
2
48,000
24
34,864,000
726.333
0.302225
non_additive
CC-BY-4.0
film_02
sfx
target_stem
2
48,000
24
34,864,000
726.333
0.495126
non_additive
CC-BY-4.0
film_02
linear_mix
input_mix_additive
2
48,000
24
34,864,000
726.333
0.891251
non_additive
CC-BY-4.0
film_02
production_mix
input_mix_production
2
48,000
24
34,864,000
726.333
0.495126
non_additive
CC-BY-4.0
film_03
dialogue
target_stem
1
48,000
24
30,088,000
626.833
0.339173
non_additive
CC-BY-4.0
film_03
music
target_stem
1
48,000
24
30,088,000
626.833
0.339173
non_additive
CC-BY-4.0
film_03
sfx
target_stem
1
48,000
24
30,088,000
626.833
0.339173
non_additive
CC-BY-4.0
film_03
linear_mix
input_mix_additive
1
48,000
24
30,088,000
626.833
0.89125
non_additive
CC-BY-4.0
film_03
production_mix
input_mix_production
1
48,000
24
30,088,000
626.833
0.31991
non_additive
CC-BY-4.0
film_04
dialogue
target_stem
2
48,000
24
9,272,000
193.167
0.482632
non_additive
CC-BY-3.0
film_04
music
target_stem
2
48,000
24
9,272,000
193.167
0.423009
non_additive
CC-BY-3.0
film_04
sfx
target_stem
2
48,000
24
9,272,000
193.167
0.825153
non_additive
CC-BY-3.0
film_04
linear_mix
input_mix_additive
2
48,000
24
9,272,000
193.167
0.89125
non_additive
CC-BY-3.0
film_04
production_mix
input_mix_production
2
48,000
24
9,272,000
193.167
0.426361
non_additive
CC-BY-3.0
film_05
dialogue
target_stem
2
48,000
24
21,823,888
454.664
0.529867
non_additive
CC-BY-SA-4.0
film_05
music
target_stem
2
48,000
24
21,823,888
454.664
0.717133
non_additive
CC-BY-SA-4.0
film_05
sfx
target_stem
2
48,000
24
21,823,888
454.664
0.712675
non_additive
CC-BY-SA-4.0
film_05
linear_mix
input_mix_additive
2
48,000
24
21,823,888
454.664
0.891251
non_additive
CC-BY-SA-4.0
film_05
production_mix
input_mix_production
2
48,000
24
21,823,888
454.664
0.683496
non_additive
CC-BY-SA-4.0
film_06
dialogue
target_stem
2
48,000
24
21,872,384
455.675
0.762867
non_additive
CC-BY-SA-4.0
film_06
music
target_stem
2
48,000
24
21,872,384
455.675
0.682863
non_additive
CC-BY-SA-4.0
film_06
sfx
target_stem
2
48,000
24
21,872,384
455.675
0.758174
non_additive
CC-BY-SA-4.0
film_06
linear_mix
input_mix_additive
2
48,000
24
21,872,384
455.675
0.891251
non_additive
CC-BY-SA-4.0
film_06
production_mix
input_mix_production
2
48,000
24
21,872,384
455.675
0.683675
non_additive
CC-BY-SA-4.0
film_07
dialogue
target_stem
2
48,000
24
5,440,512
113.344
0.619029
additive
CC-BY-SA-4.0
film_07
music
target_stem
2
48,000
24
5,440,512
113.344
0.687486
additive
CC-BY-SA-4.0
film_07
sfx
target_stem
2
48,000
24
5,440,512
113.344
0.875413
additive
CC-BY-SA-4.0
film_07
linear_mix
input_mix_additive
2
48,000
24
5,440,512
113.344
0.891251
additive
CC-BY-SA-4.0
film_08
dialogue
target_stem
2
48,000
24
5,460,032
113.751
0.528038
additive
CC-BY-SA-4.0
film_08
music
target_stem
2
48,000
24
5,460,032
113.751
0.687486
additive
CC-BY-SA-4.0
film_08
sfx
target_stem
2
48,000
24
5,460,032
113.751
0.875413
additive
CC-BY-SA-4.0
film_08
linear_mix
input_mix_additive
2
48,000
24
5,460,032
113.751
0.891251
additive
CC-BY-SA-4.0
film_09
dialogue
target_stem
2
48,000
24
72,000,000
1,500
0.89119
non_additive
CC-BY-SA-4.0
film_09
music
target_stem
2
48,000
24
72,000,000
1,500
0.165311
non_additive
CC-BY-SA-4.0
film_09
sfx
target_stem
2
48,000
24
72,000,000
1,500
0.497764
non_additive
CC-BY-SA-4.0
film_09
linear_mix
input_mix_additive
2
48,000
24
72,000,000
1,500
0.891251
non_additive
CC-BY-SA-4.0
film_09
production_mix
input_mix_production
2
48,000
24
72,000,000
1,500
0.356958
non_additive
CC-BY-SA-4.0

CineAudioDB

A real-world evaluation set for cinematic audio source separation.

CineAudioDB contains real film/animation productions with ground-truth stems (dialogue, music, sfx) for evaluating cinematic source-separation models. Unlike synthetic datasets that sum stems linearly, real productions are mixed with a non-linear mastering chain (compression, limiting, sidechain ducking, reverb), so the released stems do not generally sum to the mastered mix. To support fair evaluation under both assumptions, every production is provided in two input versions:

  • linear_mix — the three stems summed (perfectly additive: mix = dialogue + music + sfx).
  • production_mix — the real mastered stereo mix (non-additive; this is what audiences hear).

This is the real-content evaluation companion to the synthetic generation pipeline described in the accompanying paper.

At a glance

Productions 9 (7 open-source animations + 2 student production)
Stems dialogue, music, sfx (ambience merged into sfx); film_01 also ships 8 fine stems → 3-/4-stem groupings
Mix versions linear_mix (additive), production_mix (mastered, where available)
Total duration 2.46 h
Audio format FLAC (lossless), 24-bit, 48 kHz
Channels stereo (one production is mono native)

Why two mix versions

Mask-based separators (e.g. Bandit, MRX) assume mix = Σ stemₖ and estimate each source as ŝₖ = x ⊙ mₖ. Real mastered mixes break this assumption: mastering redistributes and limits energy so individual stems routinely exceed the mix in some time-frequency bins.

  • Evaluate on linear_mix to isolate model quality from the mixing-process mismatch (the input decomposes exactly into the ground-truth stems).
  • Evaluate on production_mix to measure real deployment performance on professionally mastered audio. Because the mapping is non-additive, scale-/shift-sensitive metrics (e.g. SI-SNR) are best complemented by listening tests.

Directory structure

CineAudioDB/
├── README.md
├── NOTICE.md            # source attributions (CC-BY)
├── metadata.csv         # one row per audio file (HF audiofolder-compatible)
├── films.json           # per-production summary
└── data/
    ├── film_01/         # dialogue, music, sfx, linear_mix          (additive)
    │   └── fine/        # 8 atomic print-master stems (see below)
    ├── film_02/         # dialogue, music, sfx, linear_mix, production_mix
    ├── film_03/         # dialogue, music, sfx, linear_mix, production_mix   (mono)
    └── film_04/         # dialogue, music, sfx, linear_mix, production_mix

Within a production, every file has identical length, sample rate, and channel count.

Fine stems (film_01)

film_01 is delivered as a full professional print master, so it uniquely ships its 8 atomic stem categories in data/film_01/fine/dx (dialogue), vo (voice-over), mx (music), fx (hard effects), fol (foley), bg (backgrounds/ambience), pfx (production effects), arch (archival). They are scaled by the same per-production gain as the 3-stem files, so any grouping sums exactly (to 24-bit precision) to the corresponding stem and to linear_mix:

grouping dialogue music ambience sfx
3-stem (top-level files) dx+vo mx fx+fol+bg+pfx+arch
4-stem (paper's synthesis taxonomy) dx+vo mx bg fx+fol+pfx+arch
import soundfile as sf, numpy as np
fine = "data/film_01/fine/"
def s(*names): return sum(sf.read(fine+n+".flac")[0] for n in names)
dialogue, music, ambience, sfx = s("dx","vo"), s("mx"), s("bg"), s("fx","fol","pfx","arch")
# dialogue + music + ambience + sfx == data/film_01/linear_mix.flac  (24-bit precision)

Productions (anonymized)

id duration channels mix type production_mix source license
film_01 78.00 min stereo (downmix from 5.1) additive¹ student production CC-BY-SA-4.0
film_02 12.11 min stereo non-additive open-source animation CC-BY-4.0
film_03 10.45 min mono non-additive open-source animation CC-BY-4.0
film_04 3.22 min stereo (DAW reconstruction) non-additive open-source animation CC-BY-3.0
film_05 7.58 min stereo (Blender VSE render) non-additive open-source animation (EN) CC-BY-SA-4.0
film_06 7.59 min stereo (Blender VSE render) non-additive open-source animation (RU) CC-BY-SA-4.0
film_07 1.89 min stereo (Blender VSE render) additive³ open-source animation (EN) CC-BY-SA-4.0
film_08 1.90 min stereo (Blender VSE render) additive³ open-source animation (RU) CC-BY-SA-4.0
film_09 25.00 min stereo non-additive student production CC-BY-SA-4.0

Audio format & normalization

  • FLAC (lossless), 24-bit, 48 kHz throughout — bit-exact to the source PCM (verified maxdiff=0), decoded transparently by soundfile and 🤗 datasets.
  • Surround sources are downmixed to stereo via ITU-R BS.775 (Lt = L + 0.707·C + 0.707·Ls, Rt = R + 0.707·C + 0.707·Rs; LFE dropped).
  • A single per-production gain is applied uniformly to all files of that production so the loudest signal peaks at −1 dBFS (no clipping in fixed-point). This preserves the additive property (linear_mix = g·(dialogue+music+sfx)) and the relative balance between stems and the production mix. The applied gain is recorded in films.json.
  • Additivity verified: max|dialogue+music+sfx − linear_mix| ≈ 2·10⁻⁷ (24-bit quantization floor).

Loading

import soundfile as sf

# additive evaluation: input decomposes exactly into the targets
mix,  sr = sf.read("data/film_02/linear_mix.flac")
dia,  _  = sf.read("data/film_02/dialogue.flac")
mus,  _  = sf.read("data/film_02/music.flac")
sfx,  _  = sf.read("data/film_02/sfx.flac")
# mix == dia + mus + sfx  (to 24-bit precision)

# real-world evaluation: mastered, non-additive input
prod, _  = sf.read("data/film_02/production_mix.flac")

Or via 🤗 datasets (each row = one audio file; filter on component / kind / film_id):

from datasets import load_dataset
ds = load_dataset("audiofolder", data_dir="CineAudioDB", split="test")
stems = ds.filter(lambda r: r["kind"] == "target_stem")

metadata.csv columns: file_name, film_id, component, kind, channels, sample_rate, bit_depth, frames, duration_sec, peak, mix_type, source_type, license.

Processing & provenance

Each production arrived in a different state; all were normalized to the layout above.

  • film_01 — 8 professional print-master stem categories (DX, VO, MX, FX, FOL, BG, PFX, ARCH), each as six 24-bit mono channels of a 5.1 mix. Each category was downmixed to stereo (BS.775) and grouped: dialogue = DX+VO, music = MX, sfx = FX+FOL+BG+PFX+ARCH.
  • film_02 — stems trimmed of 11.45 s trailing silence to align with the mastered mix; mono center dialogue duplicated to L=R; original 24-bit stereo master used as production_mix.
  • film_03 — native mono, 16-bit; kept mono (stereo cannot be honestly recovered) and upcast to 24-bit. The supplied mixture is the mastered production_mix.
  • film_04 — reconstructed from a DAW (Ardour) session by rendering each region (source slice, region gain, fades, center pan) into per-category stereo buses; reverb sends and master-bus plugins are intentionally not replicated, which is what makes the stems dry and the mix non-additive. The real session render is the production_mix, advanced by 682 samples (~14 ms) to compensate for master-bus latency and align with the stems.

Known limitations

  • film_03 is mono. Its dialogue/music/sfx and both mixes are single-channel.
  • film_04 stems are dry reconstructions. They omit master-bus reverb/EQ/limiting; the intro/outro theme music audible in its production_mix is not isolated in the stems (it existed only inside a full-mix bounce and was excluded to keep the music target clean). This is a genuine non-additive gap.
  • production_mix ≠ Σ stems for film_02film_04 by design. Use linear_mix when you need exact additivity.
  • Surround productions lose the LFE channel in the stereo downmix (standard practice).

Citation

@inproceedings{cineaudiodb2026,
  title     = {Realistic Data Generation and Evaluation for Cinematic Audio Source Separation},
  author    = {Anonymous},
  booktitle = {Interspeech (under review)},
  year      = {2026}
}

Licensing & attribution

This dataset mixes licenses. See NOTICE.md for the required attributions of the Creative Commons source works. The open-source animations are derived from CC-BY films (© Blender Foundation); the student productions are released under CC-BY-SA 4.0. You must comply with the per-production licenses listed in metadata.csv / films.json.

Downloads last month
159