0x3/wham

---
license: cc-by-nc-4.0
dataset_info:
features:
- name: audio
dtype: audio
- name: name
dtype: string
---

# WHAM!48kHz noise dataset

This is a mirror of the [WHAM!48kHz noise dataset](http://wham.whisper.ai/).
The original files were segmented and converted from WAV to Opus to reduce the size and accelerate streaming.

- **Sampling rate**: 48 kHz
- **Channels**: 2
- **Format**: Opus
- **Splits**:
- **Train**: 59 hours, 21216 segments, files 000 to 188
- **Validation**: 12 hours, 4444 segments, files 189 to 225
- **Test**: 7 hours, 2613 segments, files 226 to 249
- **License:** CC BY-NC 4.0
- **Source:** [http://wham.whisper.ai/](http://wham.whisper.ai/)
- **Paper:** [WHAM!: Extending Speech Separation to Noisy Environments](https://arxiv.org/abs/1907.01160)

## Usage

```python
import io

import soundfile as sf
from datasets import Features, Value, load_dataset

for item in load_dataset(
"philgzl/wham",
split="train",
streaming=True,
features=Features({"audio": Value("binary"), "name": Value("string")}),
):
print(item["name"])
buffer = io.BytesIO(item["audio"])
x, fs = sf.read(buffer)
# do stuff...
```

## Citation

```bibtex
@inproceedings{wichern2019wham,
title = {{WHAM!}: {Extending} speech separation to noisy environments},
author = {Wichern, Gordon and Antognini, Joe and Flynn, Michael and Zhu, Licheng Richard and McQuinn, Emmett and Crow, Dwight and Manilow, Ethan and Roux, Jonathan Le},
booktitle = {Proc. Interspeech},
pages = {1368--1372},
year = {2019},
}
```