humyn-labs/High-Fidelity-TTS

---
license: cc-by-4.0
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
dataset_info:
features:
- name: audio
dtype: audio
- name: file_name
dtype: string
- name: language
dtype: string
- name: gender
dtype: string
splits:
- name: train
num_bytes: 139109153
num_examples: 23
download_size: 139113103
dataset_size: 139109153
---
# High-Quality TTS Speech Dataset

This dataset contains clean, high-quality human-recorded speech clips under studio environment designed for **neural Text-to-Speech (TTS)** model training. Each recording is captured in a quiet environment with clear pronunciation and consistent pacing.

---

## Dataset Features

- Studio-quality microphone recordings
- Minimal background noise
- Consistent tone, pacing, and speaking style
- Suitable for both research and commercial TTS modeling with attribution

---

## Intended Uses

### ✅ Direct Use

- Training neural Text-to-Speech (TTS) models
- Benchmarking voice synthesis quality
- Prosody and voice-style modeling
- Multilingual and accent adaptation research
- Phoneme, grapheme, and linguistic modeling

### ❌ Out-of-Scope Use

- Real-time, mission-critical speech systems
- Medical or diagnostic speech analysis
- Commercial deployment without proper CC BY 4.0 credit
- Biometric or individual identity recognition

---

## Considerations and Limitations

- ❗ Dataset size is limited (<1,000 samples) and may not cover all phonetic diversity
- 🎧 Voice style is consistent; may not generalize to diverse accents or emotional variations
- 🔄 Future expansions will include more speakers, accents, and emotions for better generalization

---

## License

**CC BY 4.0** — Free to use, modify, distribute, and publish with attribution.

---

## Contact

For dataset-related queries, please contact:

**[[[e-mail]](mailto:[e-mail])]**