The dataset viewer is not available for this split.
Server error while post-processing the split rows. Please report the issue.
Error code: RowsPostProcessingError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Baseline Orpheus TTS Combined Dataset
Overview
This dataset contains synthetic speech generated using the unmodified Orpheus-3B model across all pathological and healthy speakers for baseline comparison.
Statistics
- Total Samples: 800
- Total Duration: 1534.04 seconds (0.4 hours)
- Average Duration: 1.92 seconds
- Number of Speakers: 8
- Sample Rate: 24,000 Hz
Corpus Breakdown
- LibriSpeech_Healthy: 60 samples, 554.4s total, 9.24s avg
- UA-Speech_Dysarthric: 400 samples, 482.2s total, 1.21s avg
- TORGO_Healthy: 200 samples, 289.8s total, 1.45s avg
- TORGO_Dysarthric: 140 samples, 207.6s total, 1.48s avg
Speaker Coverage
- TORGO Dysarthric: F04, M02
- TORGO Healthy: FC02, MC01
- UA-Speech: F02, M04
- LibriSpeech: 211, 4014
Model Information
- Base Model: unsloth/orpheus-3b-0.1-ft (unmodified)
- Purpose: Baseline comparison for pathological speech synthesis evaluation
- Generation Date: 2025-09-13T14:36:30.220298
Usage
from datasets import Dataset
# Load combined dataset
dataset = Dataset.from_parquet("baseline_orpheus_combined.parquet")
# Filter by corpus
torgo_samples = dataset.filter(lambda x: x['corpus'] == 'TORGO')
dysarthric_samples = dataset.filter(lambda x: x['condition'] == 'Dysarthric')
# Access audio and text
for sample in dataset.select(range(3)):
print(f"Speaker {sample['speaker_id']}: {sample['text']}")
audio_array = sample['audio']['array']
print(f"Audio shape: {audio_array.shape}")
- Downloads last month
- 6