DigitalUmuganda/Afrivoice_V2

Dataset Card for the image text and voice dataset

Dataset Description

This is an image prompt ASR dataset 4 African languages; Kirundi, Ndau, Ndebele, and Oshiwambo. Each language was collected on at most 2 domains.

Language	Domain	Total hours	Total transcribed hours	Total clips	Size (GB)
Kirundi	Agriculture	125.42	125.42	24,926	7.2327
	Education	383.66	383.66	72,994	24.6425
	Sub Total	509.08	509.08	97,920	31.8752
Ndau	Education	90.60	90.60	17,533	6.1026
	Health	396.97	396.97	77,202	19.3454
	Sub Total	487.57	487.57	94,735	25.4480
Ndebele	Education	101.10	101.10	18,902	6.5181
	Health	397.77	397.77	74,778	18.4713
	Sub Total	498.87	498.87	93,680	24.9894
Oshiwambo	Agriculture	255.84	255.84	44,850	15.1994
	Finance	181.98	181.98	32,055	10.8009
	Sub Total	437.82	437.82	76,905	26.0003
Total		1,933.34	1,933.34	363,240	108.3129

How to use

The huggingface_hub library allows you to load and pre-process your dataset in pure Python, at scale. The dataset can be downloaded in one call to your local drive by using the snapshow_download function.

For full dataset

from huggingface_hub import snapshot_download                                                   

snapshot_download(
    repo_id="DigitalUmuganda/Afrivoice_V2",
    repo_type='dataset',
    local_dir='<destination_dir>'
    )

Download a particular language

from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="DigitalUmuganda/Afrivoice_V2",
    repo_type="dataset",
    local_dir="<destination_dir>",
    allow_patterns="Kirundi/*"
)

The languages options are:

• Kirundi
• Ndau
• Ndebele
• Oshiwambo

Dataset Structure

Data Instance

{
    "voice_creator_id":"KGDIxAHRYAOvBRkfZGgi5708Hz82",
    "image_filepath":"6cbf90cf3951_0.jpg",
    "image_category":"agriculture",
    "image_sub_category":"Food banks & food crops and farmers",
    "category":"agriculture",
    "audio_filepath":"UKfpM5rleuyc0T85HWOn.webm",
    "transcription":"Abagabo batatu bahagaze mu murima, bikoreye ku bitugu, udukoresho babikamwo ivyo basoromye. Inyuma y'abo hakaba hari ibiti bitari bike bihatewe.",
    "normalized_transcription":"abagabo batatu bahagaze mu murima bikoreye ku bitugu udukoresho babikamwo ivyo basoromye inyuma yabo hakaba hari ibiti bitari bike bihatewe",
    "age_group":"18-24",
    "gender":"Male",
    "project_name":"Digital Umuganda",
    "locale":null,
    "year":2026,
    "duration":15.48,
    "location":null,
    "uid":"KGDIxAHRYAOvBRkfZGgi5708Hz82",
    "key":"UKfpM5rleuyc0T85HWOn",
    "dir_path":"agriculture_kirundi_dev",
    "shard_id":0,
    "image_shard_id":0
}

Data Fields

voice_creator_id (string): An id for which client (voice) made the recording

transcription_creator_id (string): An id for which client (text) made the transcription

transcription (string): Original audio transcription with punctuation and capitalization

normalized_transcription (string): Original audio transcription without punctuation and capitalization

image_filepath (string): file path of the image file inside the dataset

audio_filepath (string): file path of the audio file inside the dataset

dir_path (string): main directory path of the record

shard_id (string): chunk identification number where the audio record belong

image_shard_id (string): chunk identification number where the image record belong

age_group (string): age range of the audio recorder

gender (string): The gender of the speaker

location (string): geographical location of the audio recorder

duration (int): length in seconds of the audio file

image_category (string): domain of the image (eg: health, agriculture, finance), used as prompt during audio creation.

image_sub_category (string): Sub-domain label of the image (e.g., within agriculture: “seed farming” or “forestry”), used to guide audio creation.

category (string): category of the record

project (string): project name

locale (string): The locale of the speaker

year (int): Year of recording

project_name (string): project name

location (string): location of the project

key (string): key identifier of the data point

Licensing Information

All datasets are licensed under the Creative Commons license (CC-BY-4).