Upload 4 files

README.md

@@ -0,0 +1,139 @@
+---
+license: mit
+language: en
+tags:
+- pathological-speech
+- speech-synthesis
+- tts
+- voice-conversion
+- healthy-control
+- torgo
+---
+
+# Torgo Healthy Male Dataset
+
+## Overview
+
+This dataset contains healthy control speech samples from a male speaker (MC01) in the TORGO corpus, prepared for pathological speech synthesis research.
+
+**Speaker Information:**
+- **Speaker ID:** MC01
+- **Corpus:** TORGO
+- **Gender:** Male
+- **Speech Status:** Healthy Control
+- **Disorder Type:** None
+- **Severity:** None
+
+## Dataset Statistics
+
+- **Total Samples:** 800
+- **Total Duration:** 0.66 hours
+- **Sampling Rate:** 24,000 Hz
+- **Format:** Audio arrays with transcriptions
+
+### Training Split
+- **Samples:** 700
+- **Duration:** 0.57 hours
+- **Avg Duration:** 2.9s
+- **Duration Range:** 0.9s - 8.6s
+- **Avg Text Length:** 11 characters
+
+### Test Split  
+- **Samples:** 100
+- **Duration:** 0.09 hours
+- **Avg Duration:** 3.1s
+- **Duration Range:** 1.2s - 7.7s
+- **Avg Text Length:** 12 characters
+
+## Usage
+
+This dataset is optimized for:
+- **Text-to-Speech (TTS) Fine-tuning** with models like Orpheus-TTS
+- **Voice Conversion (VC)** experiments with kNN-VC and similar methods
+- **Hybrid TTS+VC** pipeline development
+- **ASR Evaluation** on pathological speech
+- **Speech Synthesis Research** for accessibility applications
+
+### Loading the Dataset
+
+```python
+from datasets import load_dataset
+
+# Load the dataset
+dataset = load_dataset("your-username/torgo_healthy_male")
+
+# 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
+
+```python
+from transformers import Trainer
+from datasets import load_dataset
+
+# Load and use directly with Trainer (no preprocessing needed)
+dataset = load_dataset("your-username/torgo_healthy_male")
+trainer = Trainer(
+    train_dataset=dataset['train'],
+    eval_dataset=dataset['test'],
+    # ... other trainer arguments
+)
+```
+
+## Research Applications
+
+This dataset is part of a larger study on pathological speech synthesis and ASR degradation, investigating:
+
+1. **TTS Fine-tuning**: Speaker-specific fine-tuning of large TTS models
+2. **Voice Conversion**: Zero-shot conversion from healthy to pathological speech
+3. **Hybrid Approaches**: Combining TTS content generation with VC-based acoustic styling
+4. **ASR Robustness**: Evaluating ASR performance on synthetic pathological speech
+
+## Citation
+
+If you use this dataset in your research, please cite:
+
+```bibtex
+@dataset{torgo_healthy_male_2025,
+  title={Torgo Healthy Male Dataset for Pathological Speech Synthesis},
+  author={Your Name},
+  year={2025},
+  publisher={Hugging Face},
+  url={https://huggingface.co/datasets/your-username/torgo_healthy_male}
+}
+```
+
+## Original Corpus Citation
+
+**TORGO Corpus:**
+- TORGO: Please cite the original TORGO database papers
+- UA-Speech: Please cite the original UA-Speech corpus papers  
+- LibriSpeech: Please cite the original LibriSpeech papers
+
+## License
+
+MIT License - See LICENSE file for details.
+
+## Ethical Considerations
+
+This dataset contains speech from individuals with speech disorders and should be used responsibly:
+- Respect speaker privacy and dignity
+- Use for legitimate research purposes
+- Consider potential biases in synthetic speech generation
+- Ensure accessibility improvements benefit the target communities
+
+## Contact
+
+For questions about this dataset or the associated research, please contact [your-email].

dataset_info.json

@@ -0,0 +1,45 @@
+{
+  "dataset_name": "torgo_healthy_male",
+  "speaker_info": {
+    "speaker_id": "MC01",
+    "corpus": "TORGO",
+    "gender": "Male",
+    "speech_status": "Healthy Control",
+    "disorder_type": "None",
+    "severity": "None"
+  },
+  "train_stats": {
+    "samples": 700,
+    "total_duration_hours": 0.57,
+    "avg_duration_seconds": 2.94,
+    "min_duration_seconds": 0.92,
+    "max_duration_seconds": 8.55,
+    "avg_text_length": 11.3,
+    "sampling_rate": 24000
+  },
+  "test_stats": {
+    "samples": 100,
+    "total_duration_hours": 0.09,
+    "avg_duration_seconds": 3.08,
+    "min_duration_seconds": 1.2,
+    "max_duration_seconds": 7.65,
+    "avg_text_length": 12.4,
+    "sampling_rate": 24000
+  },
+  "total_samples": 800,
+  "total_duration_hours": 0.6599999999999999,
+  "sampling_rate": 24000,
+  "format": "parquet",
+  "columns": [
+    "audio",
+    "text"
+  ],
+  "creation_date": "2025-09-10T23:19:44.111519",
+  "compatible_with": [
+    "huggingface",
+    "transformers",
+    "unsloth",
+    "orpheus-tts",
+    "knn-vc"
+  ]
+}

test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:342b3d913b021da4a0e58ae97f8306b916310f830d0b8fb5b70c515d525d3254
+size 14761988

train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1f120ee773113e5c82489a3122a2dfcdc5a0a2eceaba3ce6792c9fceb9c0dd2
+size 98475867