README.md

metadata

language:
  - en
license: mit
multimodal: true
task_categories:
  - audio-to-text
  - text-to-audio

SNAC Encoded Elise Dataset

This dataset contains audio samples from the Elise dataset encoded using the SNAC (Sound Neural Audio Codec) model.

Dataset Description

• Source: MrDragonFox/Elise dataset
• Encoding: SNAC audio codec (hubertsiuzdak/snac_24khz)
• Sampling Rate: 24kHz
• Format: Text transcriptions paired with SNAC discrete codes

Structure

Each sample contains:

• text: Text transcription
• snac_codes: Multi-layer SNAC audio codes
• audio_length_seconds: Duration of original audio
• original_sample: Metadata from source dataset

Usage

from datasets import load_dataset

dataset = load_dataset("{repo_name}")

Citation

If you use this dataset, please cite the original Elise dataset and the SNAC model.