Valexiem/mio_dataset

Mio Dataset

Dataset Description

This dataset contains voice samples of the character Mio from the Crush Crush / Hush Hush games. It consists of approximately 6,000 audio files with corresponding transcriptions, suitable for text-to-speech (TTS) model training and voice cloning applications.

Dataset Structure

The dataset is organized as follows:

• wavs/ - Directory containing 6,036 audio files in WAV format
• metadata.csv - CSV file with two columns:
  • file_name: Path to the audio file
  • transcription: Text transcription of the audio

Example Entry

file_name,transcription
wavs/mio_dataset_00001.wav,"No, does not want."
wavs/mio_dataset_00002.wav,"You should feel lucky."

Usage

This dataset can be used for:

• Training TTS models
• Voice cloning experiments
• Character voice synthesis
• Speech recognition fine-tuning

Licensing

Please ensure you have the appropriate rights and permissions before using this dataset for commercial purposes. This dataset contains voice recordings from a commercial game property.

Citation

If you use this dataset in your research or projects, please provide appropriate attribution to the original game developers.