README.md

metadata

license: other
configs:
  - config_name: hau
    data_files:
      - split: train
        path: hau/train-*
      - split: dev
        path: hau/dev-*
      - split: test
        path: hau/test-*
      - split: rejected
        path: hau/rejected-*
      - split: pending
        path: hau/pending-*
  - config_name: kau
    data_files:
      - split: train
        path: kau/train-*
      - split: dev
        path: kau/dev-*
      - split: test
        path: kau/test-*
      - split: rejected
        path: kau/rejected-*
      - split: pending
        path: kau/pending-*
  - config_name: shu
    data_files:
      - split: train
        path: shu/train-*
      - split: dev
        path: shu/dev-*
      - split: test
        path: shu/test-*
      - split: rejected
        path: shu/rejected-*
      - split: pending
        path: shu/pending-*
dataset_info:
  - config_name: hau
    features:
      - name: id
        dtype: string
      - name: path
        dtype:
          audio:
            sampling_rate: 48000
      - name: sentence
        dtype: string
      - name: task_id
        dtype: int64
      - name: sentence_source
        dtype: string
      - name: user_id
        dtype: string
      - name: age
        dtype: string
      - name: gender
        dtype: string
      - name: duration
        dtype: float64
      - name: locale
        dtype: string
      - name: variant
        dtype: string
      - name: education_level
        dtype: string
      - name: country_of_origin
        dtype: string
      - name: created_at
        dtype: string
    splits:
      - name: train
        num_bytes: 14827524621.128
        num_examples: 26996
      - name: dev
        num_bytes: 1653234645.88
        num_examples: 2540
      - name: test
        num_bytes: 2887837588.52
        num_examples: 4591
      - name: rejected
        num_bytes: 782755135.265
        num_examples: 1265
      - name: pending
        num_bytes: 537930413.212
        num_examples: 1273
    download_size: 16023003452
    dataset_size: 20689282404.005
  - config_name: kau
    features:
      - name: id
        dtype: string
      - name: path
        dtype:
          audio:
            sampling_rate: 48000
      - name: sentence
        dtype: string
      - name: task_id
        dtype: int64
      - name: sentence_source
        dtype: string
      - name: user_id
        dtype: string
      - name: age
        dtype: string
      - name: gender
        dtype: string
      - name: duration
        dtype: float64
      - name: locale
        dtype: string
      - name: variant
        dtype: string
      - name: education_level
        dtype: string
      - name: country_of_origin
        dtype: string
      - name: created_at
        dtype: string
    splits:
      - name: train
        num_bytes: 11196364730.204
        num_examples: 17149
      - name: dev
        num_bytes: 2109040286.8
        num_examples: 3485
      - name: test
        num_bytes: 2079799640.24
        num_examples: 3865
      - name: rejected
        num_bytes: 2459403718.264
        num_examples: 3044
      - name: pending
        num_bytes: 127497059
        num_examples: 180
    download_size: 14242864227
    dataset_size: 17972105434.508
  - config_name: shu
    features:
      - name: id
        dtype: string
      - name: path
        dtype:
          audio:
            sampling_rate: 48000
      - name: sentence
        dtype: string
      - name: task_id
        dtype: int64
      - name: sentence_source
        dtype: string
      - name: user_id
        dtype: string
      - name: age
        dtype: string
      - name: gender
        dtype: string
      - name: duration
        dtype: float64
      - name: locale
        dtype: string
      - name: variant
        dtype: string
      - name: education_level
        dtype: string
      - name: country_of_origin
        dtype: string
      - name: created_at
        dtype: string
    splits:
      - name: train
        num_bytes: 4009726536.84
        num_examples: 5702
      - name: dev
        num_bytes: 429643651
        num_examples: 657
      - name: test
        num_bytes: 429377999
        num_examples: 718
      - name: rejected
        num_bytes: 95954787
        num_examples: 134
      - name: pending
        num_bytes: 679906469.768
        num_examples: 1034
    download_size: 4083273280
    dataset_size: 5644609443.608
task_categories:
  - automatic-speech-recognition
language:
  - ha
  - kr
  - ar
size_categories:
  - 10K<n<100K
extra_gated_prompt: >-
  :warning: This dataset is subject to access controls. To download the data,
  you must request access and accept the [Dataset Access
  Terms](https://huggingface.co/datasets/CLEAR-Global/TWB-Voice-dataset-access-terms).
  After approval, the dataset is provided under the terms of CC BY-NC 4.0. Note:
  These additional conditions mean this dataset is not a fully open dataset
  under Creative Commons licensing alone.

  By checking the box below, you confirm that: 

  - You read and agree to the terms outlined in the Dataset Access Terms. 

  - You understand that by downloading the dataset, you become a data
  controller, responsible for complying with applicable data protection laws,
  including acting on any data subject rights requests (such as deletion or
  correction). 

  - If you redistribute this dataset, you will ensure that the recipients will
  use it under the same license and Dataset Access Terms.

  - In the event the dataset is updated in the future or if CLEAR Global
  receives a data deletion request, we will notify all registered downloaders.

  It is your responsibility as a data controller to implement such updates or
  deletions in your environment and other datasets that you might have
  distributed.
extra_gated_fields:
  I have read and agree to the Dataset Access Terms: checkbox
  Would you be willing to be contacted by email in the future to provide brief feedback on how you have used this dataset?:
    type: select
    options:
      - Yes, I agree to be contacted for feedback
      - No, I do not wish to be contacted

TWB Voice Dataset v1.0

Dataset Summary

TWB Voice 1.0 is a multilingual speech corpus containing read speech data in three languages from Nigeria: Hausa, Shuwa Arabic, and Kanuri. This dataset was created as part of the TWB Voice project by CLEAR Global (formerly Translators without Borders) to support automatic speech recognition (ASR) development for underrepresented languages.

Languages

• Hausa (hau): Major language spoken in Nigeria, Niger, and neighboring regions. Also widely used as a lingua franca across West Africa
• Shuwa Arabic (shu): Variety of Arabic spoken by the Shuwa Arab people in the Lake Chad region
• Kanuri (kau): Language spoken around Lake Chad in Nigeria, Niger, Chad, and Cameroon

Supported Tasks

• Automatic Speech Recognition (ASR): Primary intended use case. Our baseline models can be accessed at:
  • Hausa Whisper-small fine-tuned
  • Hausa Wav2Vec-Bert-2.0 fine-tuned
  • Kanuri Whisper-small fine-tuned
  • Kanuri Wav2Vec-Bert-2.0 fine-tuned
• Speaker Recognition: Demographic metadata included for speaker analysis
• Language Identification: Multi-language corpus suitable for language ID tasks
• Speech Synthesis: Can also be used for TTS model training but you can also check our TTS-focused datasets for
  • Hausa
  • Kanuri

Dataset Structure

Data Instances

Each data instance contains:

• Audio file path and audio data
• Sentence text
• Speaker demographic information (age range, gender, education level, country of origin)
• Recording metadata

Data Fields

• id: Unique recording identifier
• path: Audio file path
• sentence: Read text prompt
• task_id: Task identifier for the recorded text prompt
• sentence_source: Source of the text content
• user_id: Speaker identifier
• age: Speaker age category (teens (18+), twenties, thirties, forties, fifties, sixties, seventies, over eighty)
• gender: Speaker gender
• duration: Audio duration in seconds
• locale: Language code
• variant: Language variant/dialect
• education_level: Speaker’s education level
• country_of_origin: Speaker’s country of origin
• created_at: Recording timestamp

Data Splits

Each language configuration contains:

• train: Training set (~80% of approved recordings)
• dev: Development/validation set (~10% of approved recordings)
• test: Test set (~10% of approved recordings)
• rejected: Recordings that failed quality review
• pending: Recordings awaiting quality review

Splits are created to ensure no speaker overlap between train/dev/test sets while maintaining approximately 80/10/10 duration ratios.

Dataset Statistics

Hours summary by language and split:

	train	dev	test	total approved	pending	rejected	total collected
hau	43.79	4.03	6.56	54.38	1.48	2.25	58.11
kau	33.18	6.27	5.11	44.56	0.37	7.04	51.97
shu	10.03	1.24	1.24	12.51	1.97	0.28	14.76
TOTAL	87.00	11.54	12.91	111.45	3.81	9.57	124.84

Hours approved by gender and language:

	male	female	total
hau	38.82	15.56	54.38
kau	22.54	22.02	44.56
shu	12.40	0.12	12.51
TOTAL	73.76	37.70	111.45

Usage

Loading the Dataset

from datasets import load_dataset

# Load specific language
hausa_dataset = load_dataset("CLEAR-Global/twb-voice-1.0", "hau")
shuwa_dataset = load_dataset("CLEAR-Global/twb-voice-1.0", "shu")
kanuri_dataset = load_dataset("CLEAR-Global/twb-voice-1.0", "kau")

# Load specific split
train_data = load_dataset("CLEAR-Global/twb-voice-1.0", "hau", split="train")

# Stream data for large datasets
dataset = load_dataset("CLEAR-Global/twb-voice-1.0", "hau", streaming=True)

Dataset Creation

Data Collection

The dataset was collected through TWB Voice platform coordinated by CLEAR Global. Native speakers were asked to read prompted text in their respective languages.

Quality Control

• All recordings, except for those in rejected and pending splits, underwent human review by native speakers.
• Only approved recordings are included in train/dev/test splits
• Rejected recordings and those pending for review are preserved in separate splits for analysis

Ethical Considerations

• All speakers consented to data collection and open publishing
• Speaker identities are anonymized using user IDs
• No attempts should be made to identify individual speakers
• Data should be used responsibly for technology development

Citation

@dataset{twb_voice_2024,
  title = {TWB Voice 1.0},
  author = {CLEAR Global},
  year = {2024},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/CLEAR-Global/twb-voice-1.0}
}

License

The dataset is released under CC BY-NC 4.0 after approval under the Dataset Access Terms. Access is gated for compliance with privacy and data protection laws (e.g., GDPR).

If your intended use of this dataset may be commercial in nature (such as building models into services that you monetize in any way), or you are unsure whether your use complies with the non-commercial restriction, especially for initiatives related to social good or public benefit, we encourage you to contact us to discuss potential licensing options or permissions. We are open to supporting impactful uses beyond the standard license terms. Contact us at tech@clearglobal.org, including a link to the dataset in question and a brief description of your intended use.

Acknowledgments

This dataset was created by CLEAR Global with support from the Patrick J. McGovern Foundation.

Contact

For questions about this dataset, please contact CLEAR Global or open an issue in the dataset repository.