metadata

configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*
      - split: validation
        path: data/validation-*
task_categories:
  - automatic-speech-recognition
language:
  - enm
tags:
  - speech
  - audio
  - whisper
size_categories:
  - 1K<n<10K
license: cc-by-nc-4.0
dataset_info:
  features:
    - name: input_features
      sequence:
        sequence: float32
    - name: labels
      sequence: int64
  splits:
    - name: train
      num_bytes: 2932208048
      num_examples: 3053
    - name: test
      num_bytes: 163274288
      num_examples: 170
    - name: validation
      num_bytes: 163274312
      num_examples: 170
  download_size: 461253903
  dataset_size: 3258756648

Enenlhet Whisper Dataset

This dataset contains preprocessed audio features and tokenized text for training Whisper models on the Enenlhet language.

Dataset Summary

Train: 3,053 examples
Test: 170 examples
Validation: 170 examples
Total: 3,393 examples

Features

input_values: Preprocessed audio features (16kHz, normalized float32 arrays)
labels: Tokenized text as integer sequences

Usage

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("sjhuskey/enenlhet-whisper-dataset")

# Access splits
train_dataset = dataset['train']
test_dataset = dataset['test'] 
validation_dataset = dataset['validation']

Citation

If you use this dataset, please cite:

@dataset{enenlhet_whisper_dataset,
  title={Enenlhet Whisper Dataset},
  author={Samuel J. Huskey},
  year={2025},
  url={https://huggingface.co/datasets/sjhuskey/enenlhet-whisper-dataset}
}