README.md

metadata

license: cc-by-4.0
task_categories:
  - text-to-speech
  - automatic-speech-recognition
language:
  - vi
tags:
  - audio
  - tts
  - vietnamese
  - news
  - single-speaker
pretty_name: TTSDistil
size_categories:
  - 10K<n<100K
dataset_info:
  features:
    - name: text_id
      dtype: string
    - name: source
      dtype: string
    - name: raw_text
      dtype: string
    - name: normalized_text
      dtype: string
    - name: transcript
      dtype: string
    - name: normalized_transcript
      dtype: string
    - name: style
      dtype: string
    - name: speaker
      dtype: string
    - name: duration_s
      dtype: float32
    - name: audio
      dtype: audio
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/*.parquet

Overview

This repository contains a speech dataset developed for Text-to-Speech (TTS) research and model training.

The dataset is part of an ongoing research project focused on building high-quality speech corpora for modern neural TTS systems. It is actively maintained, with continuous improvements in data quality, transcription consistency, metadata, and organization.

This repository is intended to host research data used throughout the development process and is not intended to serve as a general-purpose file storage repository.

Intended Use

The dataset is intended for research and development in speech synthesis, including but not limited to:

• Neural Text-to-Speech (TTS)
• Zero-shot and few-shot TTS
• Speaker adaptation
• Voice cloning research
• Speech representation learning
• Speech foundation model training

The dataset should only be used for legitimate research and development purposes and in compliance with all applicable laws and licensing requirements.

Access Policy

This dataset is gated, and access requests are reviewed manually.

To request access, please include:

• Your affiliation (individual researcher, university, research lab, or company)
• A brief description of your project
• The intended use of the dataset
• Whether the use is for research or commercial purposes

Requests without sufficient information may not be approved.

The manual review process is intended to ensure responsible use of the dataset while the project remains under active development.

Contact

If you are interested in using this dataset, please contact me with a brief description of your intended research or application. Access requests are evaluated on a case-by-case basis.

Additional documentation will be published as the project continues to mature.