Datasets:

Veronica1NW
/

cv17_sw_kenyan_sample

dataset_info:
  features:
    - name: audio
      dtype:
        audio:
          sampling_rate: 16000
    - name: sentence
      dtype: string
    - name: client_id
      dtype: string
    - name: age
      dtype: string
    - name: gender
      dtype: string
    - name: accent
      dtype: string
    - name: locale
      dtype: string
    - name: source_path
      dtype: string
  splits:
    - name: train
      num_bytes: 275284152
      num_examples: 1500
  download_size: 263378447
  dataset_size: 275284152
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: cc0-1.0
task_categories:
  - automatic-speech-recognition
language:
  - sw
tags:
  - swahili
  - speech
  - kenyan
  - audio
  - common-voice
  - automatic-speech-recognition
pretty_name: 'Common Voice 17.0 (Swahili, Kenyan Sample) '
size_categories:
  - 1K<n<10K

Common Voice 17.0 — Swahili (Kenyan Sample)

This dataset is a filtered sample of the Mozilla Common Voice 17.0 corpus, focusing on Swahili (sw) speech with Kenyan voices.

It has been subsetted for experimentation and prototyping in ASR (Automatic Speech Recognition) models targeting speech-impaired users in Kenya, covering Kenyan English and Kiswahili.

Dataset Summary

Language: Kiswahili (Swahili, sw)
Accent/Region: Kenyan speakers
Domain: Conversational, general-purpose
Source: Mozilla Common Voice v17.0
License: CC0 1.0 (public domain)

This dataset contains audio files and their corresponding transcriptions.
Each entry includes:

audio: Audio array & path
sentence: Transcription text
gender, age, accent (where available)

Intended Use

This sample dataset is intended for:

Training and fine-tuning speech-to-text (ASR) systems for Kiswahili
Research on Kenyan accents in speech technology
Prototyping models for speech accessibility tools

It is not a full dataset, but a filtered subset suitable for testing and demonstrations.

Usage Example

from datasets import load_dataset

# Load dataset (requires auth if private)
ds = load_dataset("Veronica1NW/cv17_sw_kenyan_sample", use_auth_token=True)

print(ds)
print(ds["train"][0])

@misc{commonvoice,
  title = {Mozilla Common Voice},
  howpublished = {https://commonvoice.mozilla.org},
  year = {2017--2025}
}