Hugging Face's logo

Datasets:
toolevalxm
/
AudioGen-TTS-Synth

Dataset card Files
xet
 Community

The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

AudioGen-TTS-Synth

This dataset contains synthetic speech samples generated using the Bark model.

Data Sources

Synthesis Pipeline

We used the Bark text-to-speech model to generate diverse synthetic voice samples:

  1. Loaded transcripts from AudioGen-TTS-Raw
  2. Applied voice persona randomization
  3. Generated audio using Bark with default settings
  4. Post-processed for volume normalization

Data Format

Each sample contains:

  • sample_id: Unique identifier
  • original_transcript: Source transcript from AudioGen-TTS-Raw
  • voice_preset: Bark voice persona used
  • audio_path: Path to generated audio file

Model Information

Bark is a transformer-based text-to-audio model created by Suno AI. It can generate highly realistic speech as well as music, sound effects, and nonverbal communications.

Acknowledgements

Thanks to Suno AI for releasing the Bark model and to Mozilla Foundation for the original CommonVoice data!

License

The license for this dataset is MIT.
Downloads last month
16