ebellob/voxforge_spanish_enhanced

VoxForge Spanish Enhanced (CleanUNet + FlashSR)

Dataset Summary

This dataset is a processed and enhanced version of the Spanish subset of: VoxForge. Furthermore, as this is a personal project, we give no guarantees that the audio is completely clean from any artifacts or noise the CleanUNet model could not remove. However, we have personally tested the corpus via the fine-tuning of some SOTA speech models and the results have been satisfactory.

It has been created to support high-quality speech research, particularly for:

• Text-to-Speech (TTS)
• Automatic Speech Recognition (ASR)
• Speech enhancement and robustness studies

due to the lack of big and clean speech datasets in Castilian (Peninsular) Spanish. The original VoxForge Spanish recordings are crowdsourced, home-recorded speech donated by volunteers, which results in highly variable recording conditions, background noise, and microphone quality. This motivated the creation of a cleaned, enhanced release to make the corpus more usable for modern speech pipelines.

Find demo samples in: https://erikupv.github.io/CleanUNet-FlashSR-samples/

Main processing steps

1. Speech denoising (CleanUNet) All audio samples are processed using CleanUNet, a neural speech enhancement model, to reduce background noise and artifacts while preserving speech content.

2. Audio super-resolution (FlashSR) Enhanced audio is further processed using FlashSR to improve temporal and spectral resolution, resulting in cleaner and more detailed waveforms suitable for modern TTS and ASR pipelines.

The metadata structure is preserved from the original dataset unless stated otherwise.

---

Dataset Structure

Configurations

This dataset currently provides a single train split.

---

Dataset Features

• audio_id: Unique audio segment identifier
• audio: Enhanced audio waveform
• speaker_id: Speaker identifier
• country: Country of origin (spain)
• gender: Speaker gender (male / female)
• duration: Segment duration in seconds
• normalized_text: Normalized transcription

---

Dataset Size

• Number of examples: 16,427
• Approximate size: ~25.6 GB

---

Language

• Spanish (es) — Castilian / Peninsular Spanish only

This dataset contains exclusively Peninsular Spanish audio contributed by volunteers from Spain. It does not include Latin American or other regional Spanish variants, making it particularly suitable for research requiring a clean Peninsular Spanish corpus.

---

Intended Uses

This dataset is suitable for:

• High-quality Peninsular Spanish TTS training
• ASR model training and evaluation
• Speech enhancement and super-resolution research
• Robustness studies on enhanced speech data

---

Out-of-Scope Uses

• Speaker identification or biometric profiling
• Any attempt to deanonymize speakers
• Uses violating the original VoxForge GPL license or contributor terms

---

Processing Notes

• Audio waveforms are not bit-identical to the original VoxForge release.
• Speech content and textual annotations are unchanged.
• Processing may introduce minor spectral or temporal artifacts inherent to neural enhancement models.
• The original dataset does not provide quality scores (e.g., mean_quality); therefore, no quality-based filtering was applied prior to enhancement.

---

License

This dataset inherits the license of the original VoxForge corpus:

• GNU General Public License (GPL)

Please refer to the VoxForge website and original project page for exact terms.

---

Citation

If you use this dataset, please cite both the original VoxForge project and this processed version.

Original dataset

@misc{voxforge,
  title = {VoxForge: Free Speech Recognition},
  author = {{VoxForge Contributors}},
  year = {2006--2026},
  url = {https://www.voxforge.org}
}

This dataset

@dataset{voxforge_spanish_enhanced,
  title = {VoxForge Spanish Enhanced (CleanUNet + FlashSR)},
  author = {Erik Beltrán},
  year = {2026},
  note = {Enhanced with CleanUNet denoising and FlashSR super-resolution}
}

Tools used

• CleanUNet — Kong et al., "Speech Denoising in the Waveform Domain with Self-Attention", ICASSP 2022.
  Paper: arXiv:2202.07790 · Code: NVIDIA/CleanUNet

• FlashSR — Im & Nam, "FlashSR: One-step Versatile Audio Super-resolution via Diffusion Distillation", 2025.
  Paper: arXiv:2501.10807 · Code: ysharma3501/FlashSR

---

Acknowledgements

• VoxForge contributors for donating their speech recordings
• CleanUNet authors (Kong, Ping, Dantrey & Catanzaro)
• FlashSR authors (Jaekwon Im & Juhan Nam)