Datasets:

Professor
/

dagbani-speech-data

language:
  - dag
license: cc-by-4.0
multilinguality: monolingual
task_categories:
  - text-to-speech
  - automatic-speech-recognition
tags:
  - dagbani
  - dagbanli
  - speech
  - tts
  - asr
  - african-languages
  - low-resource
pretty_name: Dagbani Speech Data (Pooled)
size_categories:
  - 10K<n<100K

Dagbani Speech Data (Pooled)

A ~96.3-hour Dagbani (Dagbanli) speech corpus, drawn from a single source (WAXAL) and filtered to only genuinely transcribed audio. Part of the AfroNet multi-language TTS data effort.

Source

WAXAL (google/WaxalNLP), dag_asr config — crowdsourced, image-prompted speech (a shared collection pipeline also used for Dagaare, Ikposo, and Akan's aka_asr in this collection). 17,818 clips, 96.3h, source = waxal.

A known upstream bug, verified not to affect this release: WAXAL's dag_asr and dga_asr (Dagaare) configs have swapped audio filenames — every clip in dag_asr carries a filename prefixed with Dagaare's naming convention (dga_gh_image_...) and vice versa. This does not affect the actual audio/transcript pairing used here (each config's own rows are pulled together correctly regardless of filename content), and independent linguistic analysis of the transcript text confirmed the language labels themselves are correct: this config's text consistently uses Dagbani's documented post-nominal determiner pattern (ŋɔ/maa, e.g. "doo ŋɔ maa"), distinct from Dagaare's postverbal factive particle la found in the dga_asr config. The bug is cosmetic (filenames only), not a language mislabel.

train+validation+test splits are pooled together (intentional, same policy already applied to WAXAL's _tts configs used elsewhere in AfroNet); the unlabeled split (untranscribed audio) is excluded, consistent with AfroNet's no-auto-transcription policy.

All audio is standardized to 16 kHz mono FLAC (lossless), 1–30 second clips.

Format

The dataset ships as WebDataset-style tar shards (shards/shard-00000.tar …, ~1 GB each, one {key}.flac file per clip) plus a single manifest (manifest.parquet / manifest.jsonl):

Column	Description
`key`, `shard`	which tar file + entry holds this clip's audio
`text`	transcript (native script)
`duration`	seconds
`source`	always `waxal`
`dataset_id`	always `0`
`split`	`train` / `val` (250 clips held out for evaluation)
`speaker_id`	source-provided speaker ID
`gender`	speaker metadata where available
`domain`	always `dag_asr`
`dbfs`, `clip_ratio`, `sil_ratio`	cheap DSP quality proxies: loudness, fraction of clipped samples, fraction of near-silent frames
`has_disfluency`	always `false`

Usage

from huggingface_hub import hf_hub_download
import pandas as pd, tarfile, io, soundfile as sf

mp = hf_hub_download("Professor/dagbani-speech-data", "manifest.parquet", repo_type="dataset")
df = pd.read_parquet(mp)

row = df.iloc[0]
shard_path = hf_hub_download("Professor/dagbani-speech-data", f"shards/{row.shard}", repo_type="dataset")
with tarfile.open(shard_path) as tar:
    audio_bytes = tar.extractfile(f"{row.key}.flac").read()
arr, sr = sf.read(io.BytesIO(audio_bytes))

The tar shards are also directly readable by the webdataset library for streaming training pipelines.

Intended use & limitations

Built for Dagbani TTS/ASR research, in particular as finetuning data for a multilingual TTS model that doesn't natively support Dagbani. Speech is crowdsourced and image-prompted, with many speakers and variable recording conditions rather than studio-controlled. This is a research aggregation; usage should respect WAXAL's own terms.

License

CC BY 4.0, per the upstream WAXAL release.

Acknowledgments

Deep thanks to the WAXAL project (Google) and its Dagbani-speaking contributors for the source corpus.

This dataset was pooled by Victor Olufemi and LyngualLabs as part of the AfroNet multi-language TTS data effort.