File size: 6,977 Bytes
b67b162
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
88dc95a
b67b162
88dc95a
b67b162
88dc95a
b67b162
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
88dc95a
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
---
{
  "pretty_name": "CHIME — CompA (prepared candidate)",
  "license": "other",
  "tags": [
    "audio",
    "webdataset",
    "chime",
    "prepared-candidate"
  ],
  "configs": [
    {
      "config_name": "default",
      "data_files": [
        {
          "split": "test",
          "path": "data/test/*.tar"
        }
      ]
    }
  ],
  "license_name": "original-source-terms",
  "license_link": "https://github.com/Sreyan88/CompA"
}
---

# CHIME · CompA

This is a source-preserving **prepared candidate**, not a scientifically admitted CHIME benchmark release. It repackages the selected upstream release into split-specific WebDataset TAR shards with native audio and per-example JSON. A Parquet index permits metadata filtering without downloading audio shards.

## Source and selected scope

- Upstream: [CompA](https://github.com/Sreyan88/CompA).
- Pinned source release: `ICLR2024 author original CompA-order and CompA-attribute ZIPs, SHA256 snapshot2026-10-28`.
- Selection: Both original author-linked public ZIPs downloaded anonymously and validated; stable direct download URLs checked. Native gold subset has400 order rows and197 attribute rows, with complete actual-file joins.
- Original license/access conditions: No separate dataset license notice found in original ZIPs or repository README; public acquisition is available. Preserve the author citation and original source metadata; this acquisition recipe does not confer redistribution rights. Public author-linked Google Drive downloads; no account, access application or agreement form. Standard public virus-scan download interstitial followed; direct confirm=t URLs verified.
- Source notes: Source hashes are locally observed from authoritative anonymous HTTPS downloads, not claimed to be publisher-supplied checksums. Reuse staged bootstrap objects. Both direct URLs returned application/octet-stream with exact Content-Length.

Original authors, licensors and recording contributors retain their rights. This preparation was assembled for public noncommercial research access; this repository does not grant broader rights than the original sources. Preserve per-recording attribution and notices. `provenance/` contains original annotations/docs/licenses, source provenance and the redistribution audit. FSD50K/Clotho audio has per-recording terms; Clotho captions retain the complete custom Tampere experimental/noncommercial license. The CHIME code license does not apply to these data.

## Representation and preservation

There are **1,226 audio assets**, 571,828,314 stored-audio bytes and 571,771,878 decoded-PCM-equivalent bytes. Native sample rate, duration and channel count are retained. No loudness normalization, silence trimming, resampling or downmixing is applied.

Each sample retains its native audio codec (WAV/FLAC/MP3/OGG or another supported native container). Each payload is accompanied by `<key>.json` in an ordinary TAR. The JSON contains `split`, stored audio metadata, provenance, source groups and source targets. `manifest.jsonl` also records shard/member offsets and checksums. `metadata.parquet` provides audio and shard columns, common source/group/license columns, and exact arbitrary annotations in `metadata_json` (decode with `json.loads`). It does not duplicate audio payloads. Keeping heterogeneous annotations as JSON preserves empty objects and differing target schemas without Arrow type coercion. `release.json` hashes the audio shards and record manifest, and `publication.json` inventories all published files including attribution assets.

CHiME-3 background recordings interleave six synchronized native mono PCM16 channels into one six-channel PCM16 WAV per session, with exact decoded-sample checks and an explicit channel map. The excluded CHiME speech corpus is not included. Other audio payloads are copied byte for byte.

## Source splits

- `test`: 1,226 original audio assets.

These are original source partitions or explicitly `unspecified`/`background` partitions. ESC-50 retains five official folds; no fixed benchmark train/test assignment is invented. VocalSet's split is the supplied technique-classification singer protocol, not a universal task protocol. Background noise is not converted into artificial word or silence examples. Linked views/channels share provenance groups.

## Scientific limitations

Scientific admission remains **pending**. Native file integrity and packaging checks do not establish label correctness, perceptual deduplication, contamination freedom or benchmark validity. Existing upstream annotation errata and ambiguity are retained/documented. The source audit found declared Freesound ID overlaps within some official folds/partitions and across ESC-50, FSD50K and Clotho; FSD50K train/validation also share uploaders. Use the supplied source IDs, group metadata and `provenance/source-audit.json` when defining an evaluation protocol. Do not claim these candidate splits are universally leakage-free. Scores from this repository alone are not official CHIME scores.

## Loading

Pin an immutable Hub commit revision in experiments. A generic Hugging Face WebDataset loader can read the shard configurations:

```python
from datasets import load_dataset
samples = load_dataset("ltuncay/chime-compa", revision="<commit-sha>", streaming=True)
```

For local source-native decoding and checksum verification, use CHIME's `iter_release(path, split=..., verify_hashes=True)`. Its audio arrays have shape `[frames, channels]`; the evaluated model owns its preprocessing. Download only selected shards/index files when appropriate; benchmark execution must not silently fetch data.

## Provenance and citation

Cite the upstream dataset and recording authors as specified in the original documentation under `provenance/`. This preparation does not claim authorship of the source recordings or annotations. Exact upstream objects, checksums, release selection, preparation metadata and source URLs are recorded in `release.json`. Source Zenodo metadata/citations are retained when available. See the CHIME project implementation brief and scientific paper for the evolving evaluation protocol.

## CHIME v1 benchmark usage

> **Do not train on CHIME evaluation data:** 1294 records in the release split(s) `test`, listed one key per line in `provenance/benchmark-v1/eval-keys.txt`. Models submitted to CHIME must not see these recordings (or their source recordings) during pretraining or fine-tuning.

| Task | Family | Metric | Folds | Train | Validation | Test (eval) |
|---|---|---|---|---|---|---|
| `compa-caption-choice` | qa | exact_match | 1 | — | — | test |

Exact per-record partition assignments (including records excluded from training because they share a source with evaluation data, and digitally silent assets) are in `provenance/benchmark-v1/<task>.jsonl`; the frozen task definitions are in `provenance/benchmark-v1/usage.json`. Splits never depend on the evaluation seed.