resproj007/torgo_dysarthric_female

Torgo Dysarthric Female Dataset

Overview

This dataset contains dysarthric speech samples from a female speaker (F04) in the TORGO corpus, prepared for pathological speech synthesis research.

Speaker Information:

• Speaker ID: F04
• Corpus: TORGO
• Gender: Female
• Speech Status: Dysarthric

Dataset Statistics

• Total Samples: 670
• Total Duration: 0.60 hours
• Sampling Rate: 24,000 Hz
• Format: Audio arrays with transcriptions

Training Split

• Samples: 600
• Duration: 0.54 hours
• Avg Duration: 3.3s
• Duration Range: 0.8s - 10.2s
• Avg Text Length: 13 characters

Test Split

• Samples: 70
• Duration: 0.06 hours
• Avg Duration: 3.3s
• Duration Range: 0.6s - 10.3s
• Avg Text Length: 12 characters

Loading the Dataset

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("your-username/torgo_dysarthric_female")

# Access train and test splits
train_data = dataset['train']
test_data = dataset['test']

# Each sample contains:
# - 'audio': {'array': numpy_array, 'sampling_rate': 24000}
# - 'text': str (normalized transcription)

# Example usage
sample = train_data[0]
audio_array = sample['audio']['array']
transcription = sample['text']
sampling_rate = sample['audio']['sampling_rate']

Direct Training with Transformers

from transformers import Trainer
from datasets import load_dataset

# Load and use directly with Trainer (no preprocessing needed)
dataset = load_dataset("your-username/torgo_dysarthric_female")
trainer = Trainer(
    train_dataset=dataset['train'],
    eval_dataset=dataset['test'],
    # ... other trainer arguments
)