README.md

metadata

language:
  - lg
subsets:
  - train
  - eval
  - test

Luganda Speech Dataset

Dataset Description

This dataset contains speech recordings in Luganda, split into training, evaluation, and test sets.

Dataset Structure

The dataset is organized as follows:

• train: Training set
• eval: Evaluation set
• test: Test set

Each entry includes the following features:

• sentence: The spoken sentence
• language: The language of the recording (Luganda)
• contributor_id: ID of the contributor
• gender: Gender of the contributor
• age_group: Age group of the contributor
• voice_clip: Path to the audio file
• duration: Duration of the audio file
• up_votes: Number of up votes
• down_votes: Number of down votes
• Region: Region of the contributor
• path: Path to the audio file