README.md

metadata

language:
  - en
  - pcm
license: cc-by-nc-4.0
task_categories:
  - automatic-speech-recognition
pretty_name: PidginMed-ASR
size_categories:
  - 1K<n<10K
dataset_info:
  features:
    - name: audio
      dtype: string
    - name: transcription
      dtype: string
    - name: english_translation
      dtype: string
    - name: speaker_id
      dtype: string
    - name: duration
      dtype: float64
    - name: split
      dtype: string
  splits:
    - name: train
      num_bytes: 2041179
      num_examples: 2982
    - name: validation
      num_bytes: 788013
      num_examples: 994
    - name: test
      num_bytes: 1199738
      num_examples: 1494
  download_size: 2148333
  dataset_size: 4028930
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*

PidginMed-ASR

PidginMed-ASR is a Nigerian Pidgin English medical speech benchmark for automatic speech recognition.

Dataset Structure

The dataset contains audio recordings paired with Nigerian Pidgin transcriptions and English source translations.

Column	Description
audio	Path to audio file
transcription	Nigerian Pidgin transcript
english_translation	Original English text
speaker_id	Anonymized speaker ID
duration	Audio duration
split	Train, validation, or test split

Splits

The dataset uses a speaker-independent split.

• Train: Speaker_2, Speaker_4, Speaker_5, Speaker_7, Speaker_9, Speaker_11
• Validation: Speaker_3, Speaker_8
• Test: Speaker_1, Speaker_6, Speaker_10

License

CC BY-NC 4.0