beyoru/vietnamese-voice

Vietnamese and English Voice Transcript Dataset

Dataset Summary

This dataset contains Vietnamese and English speech audio samples paired with text transcriptions. It is designed for speech-related tasks such as Automatic Speech Recognition (ASR), speech-to-text alignment, and linguistic analysis. The dataset features a single speaker to ensure consistency in voice characteristics.

• Language: Vietnamese, English
• Size: 4480 Vietnamese samples, 3956 English samples
• Modality: Audio + Text
• Number of Speakers: 1
• Data Structure: Tabular (3 columns)

Data Schema

The dataset is organized into three columns:

Column	Type	Description
audio	string / path	File path or identifier for the audio recording (e.g., WAV, MP3).
text	string	Verbatim Vietnamese/English transcription of the spoken audio.
speaker	string	Speaker identifier. All rows contain the same single speaker ID.

Data Collection

• Speaker: One native Vietnamese speaker
• Recording Conditions: Not specified (may vary depending on collection setup)
• Content: Spoken Vietnamese/English sentences or utterances
• Transcription: Manual or semi-automatic transcription into Vietnamese/Ênglish text

Intended Use

This dataset is intended for:

• Training and evaluating Vietnamese ASR/TTS models
• Speech-to-text experiments
• Alignment between speech and text
• Research on Vietnamese phonetics or pronunciation consistency

License

This dataset is released for research, educational, and non-commercial purposes only.

Commercial use of the dataset, derived models, generated voices, or any products trained using this dataset requires explicit written permission from the dataset owner.

By using this dataset, users agree not to:

• Redistribute the dataset without permission.
• Use the dataset for commercial purposes without a commercial license.
• Impersonate the speaker or create misleading content.

For commercial licensing inquiries, please contact the dataset owner.