| --- |
| license: other |
| license_name: internal-evaluation-only |
| language: |
| - ff |
| - fuc |
| - fuf |
| - fub |
| - ffm |
| - fue |
| task_categories: |
| - automatic-speech-recognition |
| pretty_name: "INTERNAL - Pulaar ASR evaluation corpus (temporary)" |
| size_categories: |
| - 10K<n<100K |
| tags: |
| - pulaar |
| - fula |
| - internal |
| - do-not-redistribute |
| - temporary |
| dataset_info: |
| features: |
| - name: utt_id |
| dtype: string |
| - name: audio |
| dtype: audio |
| - name: text |
| dtype: string |
| - name: duration |
| dtype: float32 |
| - name: source |
| dtype: string |
| - name: source_upstream |
| dtype: string |
| - name: source_licence |
| dtype: string |
| - name: source_ref |
| dtype: string |
| - name: had_nonspeech_marker |
| dtype: bool |
| - name: in_v4_train |
| dtype: bool |
| - name: dialect |
| dtype: string |
| - name: iso_member |
| dtype: string |
| - name: speaker_id |
| dtype: string |
| - name: sex |
| dtype: string |
| - name: split |
| dtype: string |
|
|
| configs: |
| - config_name: default |
| data_files: |
| - split: train |
| path: data/train-*.parquet |
| - split: dev |
| path: data/dev-*.parquet |
| - split: test |
| path: data/test-*.parquet |
| --- |
| |
| # INTERNAL — Pulaar ASR evaluation corpus (temporary) |
|
|
| **Private, internal evaluation only. Do not redistribute, do not make public, delete after |
| the evaluation.** Two of the six upstream datasets declare no licence at all, which means |
| all rights reserved — a private copy for evaluation is defensible, a public release of the |
| same rows is not. See `LICENSE_NOTICE.md`. |
|
|
| Rebuilt from the original public sources on 2026-10-28, after the server holding the |
| previous corpus was lost. The rebuild reproduced the previous corpus row for row |
| (58983 rows, 183.49 h), so the pipeline is reproducible from a bare |
| machine and six public sources with no surviving private artefact in the chain. |
|
|
| ## Contents |
|
|
| | split | rows | hours | |
| |---|---|---| |
| | train | 54651 | 173.86 | |
| | dev | 1317 | 2.79 | |
| | test | 3015 | 6.85 | |
| | **total** | **58983** | **183.49** | |
|
|
| 463 speakers. 16 kHz mono FLAC embedded in the parquet, one utterance per |
| row, nothing shorter than 0.4 s or longer than 30 s, no empty transcripts. |
|
|
| ## How the splits were cut |
|
|
| The hold-out unit is the **whole broadcast** for `kallaama_src` and the **speaker** |
| wherever a speaker id exists, so no speaker and no recording appears in two splits |
| (verified: 0 groups span a split boundary). Dev/test utterances whose exact transcript also |
| appears in train were then purged — dev 62, test 102 — because the n-gram language model is |
| built from train text, which makes a repeated reference easier than it looks. |
|
|
| **Known limitation:** `abdouaziz_multi_dialect` ships no speaker metadata, so the seven |
| varieties it supplies cannot be speaker-disjoint. Per-variety counts are in `stats.json` |
| under `split_provenance.rows_without_speaker_metadata`. |
|
|
| Both Kallaama derivatives (`ngia/ASR_pulaar`, `Alioukun/kallaama-pulaar-filtered`) are |
| **excluded**: their recordings are exactly the 83 SLR151 broadcasts that `kallaama_src` |
| already covers from the authoritative STM annotations, so they add no new speech, and the |
| re-segmentation was the source of the cross-split contamination measured earlier |
| (7.6–9.2 WER points on radio). |
|
|
| ## Scoring pulaar_asr_v4 against this test split |
|
|
| `in_v4_train` marks every row whose exact transcript was in v4's training data. v4 predates |
| this rebuild and was cut under different split boundaries, so **32% of the test split is |
| not unseen by v4**. Filter `in_v4_train == False` for an honest number. A model trained on |
| this corpus's own `train` split needs no such filter. |
|
|
| The overlap is very uneven, so the two cleanest slices are also the two largest: |
|
|
| | variety | test rows | in v4 train | | |
| |---|---|---|---| |
| | unknown | 527 | 9 | **2%** | |
| | pulaar | 489 | 40 | **8%** | |
| | pular | 273 | 66 | **24%** | |
| | borgu | 300 | 109 | **36%** | |
| | adamawa | 229 | 103 | **45%** | |
| | bororro | 299 | 152 | **51%** | |
| | caka | 299 | 152 | **51%** | |
| | maacina | 299 | 168 | **56%** | |
| | liptako | 300 | 173 | **58%** | |
|
|
| ## Sources |
|
|
| | source | rows | upstream | licence | |
| |---|---|---|---| |
| | `abdouaziz_multi_dialect` | 19916 | [abdouaziz/fulaa_pulaar](https://huggingface.co/datasets/abdouaziz/fulaa_pulaar) | **none declared** | |
| | `batbaina_fub_agri` | 799 | [BATBAINA/fulfulde-dataset-asr-agri](https://huggingface.co/datasets/BATBAINA/fulfulde-dataset-asr-agri) | **none declared** | |
| | `guizme_adlam` | 51 | [guizme/adlam_fulfulde](https://huggingface.co/datasets/guizme/adlam_fulfulde) | cc-by-4.0 | |
| | `guizme_pulaar` | 130 | [guizme/Pulaar](https://huggingface.co/datasets/guizme/Pulaar) | cc0-1.0 | |
| | `kallaama_src` | 14441 | [Kallaama Pulaar SLR151](https://www.openslr.org/151/) | cc-by-4.0 | |
| | `waxal_ful_asr` | 23646 | [google/WaxalNLP](https://huggingface.co/datasets/google/WaxalNLP) | cc-by-sa-4.0 | |
|
|
| 20715 of 58983 rows (35%) come from |
| sources that declare no licence. That is why this repository is private. |
|
|
| ## Load |
|
|
| ```python |
| from datasets import load_dataset |
| ds = load_dataset("tekkilab/pulaar-eval-data", split="test") # splits: train / dev / test |
| ``` |
|
|
| `datasets` 4.x and 5.x decode audio through **torchcodec**, which is not a dependency of |
| `datasets` itself. Either `pip install torchcodec`, or skip it and read the FLAC bytes |
| directly — which is all a row contains anyway: |
|
|
| ```python |
| import io, soundfile as sf |
| from datasets import load_dataset, Audio |
| |
| ds = load_dataset("tekkilab/pulaar-eval-data", split="test").cast_column("audio", Audio(decode=False)) |
| y, sr = sf.read(io.BytesIO(ds[0]["audio"]["bytes"]), dtype="float32") # 16 kHz mono |
| ``` |
|
|
| Every field is documented in `sources.json`, and every filter and drop reason in |
| `stats.json`. |
|
|
