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
(https://huggingface.co/docs/hub/datasets-cards)
AudioGen: Speech Synthesis with Multi-Modal Learning
📑 Paper | 🌐 Project Page | 💾 Released Resources | 📦 Repo
We release the raw audio data for our processed VoxSynth dataset, adapted from the original dataset from Mozilla Foundation's Common Voice project.
The data format for each line in the audio_samples_10k.jsonl is as follows:
{
"transcript": <the text transcription of the audio>,
"speaker_info": <speaker demographic information>,
"audio_path": <path to the audio file>,
"duration_ms": <duration in milliseconds>,
"sample_rate": 16000,
"language": <language code>,
"source": <the source of the raw audio files>,
"quality_score": <audio quality assessment score>
}
Some audio samples have been filtered due to quality constraints (background noise, clipping, etc.).
*Note: Due to varying recording conditions, some transcripts may have minor alignment issues with the audio.
License
The license for this dataset is CC0-1.0.
- Downloads last month
- 12