---
tags:
- audio
task_categories:
- audio-classification
pretty_name: background
size_categories:
- n<1K
---

## Dataset Statistics

- **Total Audio Files**: 4
- **Total Classes**: 4
- **Format**: MP3

## Structure

The dataset is organized in a folder structure where audio files are in subdirectories by label, and a `metadata.jsonl` file provides the file paths, labels, and descriptions.

### Columns

- `audio`: The audio recording.
- `label`: The category label of the background noise.
- `description`: A text description of the background noise.

## Usage

```python
from datasets import load_dataset

ds = load_dataset("sdialog/foreground")
```