xuyuefan111/audioform_dataset

---
license: mit
task_categories:
- image-to-text
- text-to-image
- audio-classification
- image-classification
- tabular-classification
tags:
- audio
- image
- multimodal
- visualization
- audio-visualization
- 3d-visualization
- synthetic
- proof-of-concept
- frequency-estimation
- generative-audio
- music-visualization
---

[](https://webxos.netlify.app)
[](https://github.com/webxos/webxos)
[](https://huggingface.co/webxos)
[](https://x.com/webxos)

<div style="
background: #00FF00;
border-left: 3px solid #00FF00;
padding: 12px;
margin: 10px 0;
font-family: 'Fira Code', 'Courier New', monospace;
color: #00FF00;
border-radius: 0 4px 4px 0;
font-size: 8px;
line-height: 1.1;
max-width: 100%;
overflow: hidden;
">
<pre style="
font-size: 6px;
line-height: 1.1;
margin: 0;
padding: 0;
color: #00FF00;
letter-spacing: 0px;
word-spacing: 0px;
">
AAA UUUUUUUU UUUUUUUUDDDDDDDDDDDDD IIIIIIIIII OOOOOOOOO FFFFFFFFFFFFFFFFFFFFFF OOOOOOOOO RRRRRRRRRRRRRRRRR MMMMMMMM MMMMMMMM
A:::A U::::::U U::::::UD::::::::::::DDD I::::::::I OO:::::::::OO F::::::::::::::::::::F OO:::::::::OO R::::::::::::::::R M:::::::M M:::::::M
A:::::A U::::::U U::::::UD:::::::::::::::DD I::::::::I OO:::::::::::::OO F::::::::::::::::::::F OO:::::::::::::OO R::::::RRRRRR:::::R M::::::::M M::::::::M
A:::::::A UU:::::U U:::::UUDDD:::::DDDDD:::::DII::::::IIO:::::::OOO:::::::OFF::::::FFFFFFFFF::::FO:::::::OOO:::::::ORR:::::R R:::::RM:::::::::M M:::::::::M
A:::::::::A U:::::U U:::::U D:::::D D:::::D I::::I O::::::O O::::::O F:::::F FFFFFFO::::::O O::::::O R::::R R:::::RM::::::::::M M::::::::::M
A:::::A:::::A U:::::D D:::::U D:::::D D:::::DI::::I O:::::O O:::::O F:::::F O:::::O O:::::O R::::R R:::::RM:::::::::::M M:::::::::::M
A:::::A A:::::A U:::::D D:::::U D:::::D D:::::DI::::I O:::::O O:::::O F::::::FFFFFFFFFF O:::::O O:::::O R::::RRRRRR:::::R M:::::::M::::M M::::M:::::::M
A:::::A A:::::A U:::::D D:::::U D:::::D D:::::DI::::I O:::::O O:::::O F:::::::::::::::F O:::::O O:::::O R:::::::::::::RR M::::::M M::::M M::::M M::::::M
A:::::A A:::::A U:::::D D:::::U D:::::D D:::::DI::::I O:::::O O:::::O F:::::::::::::::F O:::::O O:::::O R::::RRRRRR:::::R M::::::M M::::M::::M M::::::M
A:::::AAAAAAAAA:::::A U:::::D D:::::U D:::::D D:::::DI::::I O:::::O O:::::O F::::::FFFFFFFFFF O:::::O O:::::O R::::R R:::::RM::::::M M:::::::M M::::::M
A:::::::::::::::::::::A U:::::D D:::::U D:::::D D:::::DI::::I O:::::O O:::::O F:::::F O:::::O O:::::O R::::R R:::::RM::::::M M:::::M M::::::M
A:::::AAAAAAAAAAAAA:::::A U::::::U U::::::U D:::::D D:::::D I::::I O::::::O O::::::O F:::::F O::::::O O::::::O R::::R R:::::RM::::::M MMMMM M::::::M
A:::::A A:::::AU:::::::UUU:::::::U DDD:::::DDDDD:::::DII::::::IIO:::::::OOO:::::::OFF:::::::FF O:::::::OOO:::::::ORR:::::R R:::::RM::::::M M::::::M
A:::::A A:::::AUU:::::::::::::UU D:::::::::::::::DD I::::::::I OO:::::::::::::OO F::::::::FF OO:::::::::::::OO R::::::R R:::::RM::::::M M::::::M
A:::::A A:::::A UU:::::::::UU D::::::::::::DDD I::::::::I OO:::::::::OO F::::::::FF OO:::::::::OO R::::::R R:::::RM::::::M M::::::M
AAAAAAA AAAAAAA UUUUUUUUU DDDDDDDDDDDDD IIIIIIIIII OOOOOOOOO FFFFFFFFFFF OOOOOOOOO RRRRRRRR RRRRRRRMMMMMMMM MMMMMMMM
</pre>
</div>

## Audioform_Dataset_v1

This dataset is the very first output from **AUDIOFORM** — a Three.js powered 3D audio visualization tool that turns audio files
into beautiful, timestamped visual frames with rich metadata. **AUDIOFORM** by webXOS is available for download in the /audioform/
folder of this repo so developers can create their own similar datasets. Audio for is a synthetic harmonic oscilator that runs in HTML,
think of it as the "Hello World" / MNIST-style dataset application for audio-to-visual multimodal machine learning.

This dataset contains **10 captured frames** from a short uploaded WAV file (played at 1× speed), together with per-frame
metadata including dominant frequency, timestamp, and capture info.

## Dataset Structure

```
audioform_dataset/
├── images/
│ ├── frame_0001.png
│ ├── frame_0002.png
│ └── ... (10 PNG frames total)
├── metadata.csv # Main metadata file (Hugging Face viewer uses this)
└── README.md
```

```
| Column | Type | Description | Example Value |
|---------------|---------|-----------------------------------------------------------------------------|-----------------------------------|
| `file_name` | string | Relative path to the visualization PNG (required by Hugging Face) | `images/frame_0001.png` |
| `frame_id` | int | Sequential frame number (0-based) | 0, 1, 2, …, 9 |
| `timestamp` | float | Time in seconds when the frame was captured from the audio | 5.365, 6.219, 9.504 |
| `frequency` | int | Dominant / main detected audio frequency at capture time (Hz) | 0 (in this tiny sample) |
| `time_scale` | int | Playback speed multiplier used during visualization | 1 |
| `capture_date`| string | UTC ISO timestamp when the frame was rendered | 2026-01-13T19:57:36.427Z |
```

See how fast a tiny diffusion model / GAN / LoRA can memorize & regenerate these exact 10 styles. Use the frames as
style references for ControlNet, IP-Adapter, or fine-tuning SD to adopt this neon 3D audio-viz aesthetic.

```
This dataset shows the **format** AUDIOFORM produces.
→ Feed it real music, voices, field recordings, synths
→ Generate 1k–100k+ frames
→ Add labels (genre, instrument, mood, multiple freq peaks…)
→ Unlock serious applications:

- Music video auto-generation
- Visual audio classifiers
- Audio-conditioned image/video generation
- Interactive music → 3D art installations
- Novel multimodal music understanding models

```
## Dataset Description

This dataset was generated using AUDIOFORM, a 3D audio visualization system.

- **Total Frames**: 10
- **Generation Date**: 2026-01-13
- **Audio Type**: Uploaded WAV File
- **Time Scaling**: 1x

## Dataset Structure

- `images/`: Contains all captured frames in PNG format
- `metadata.csv`: Contains classification data for each frame

## Metadata Columns

- `file_name`: Relative path to the image file (e.g., images/frame_0001.png) - **REQUIRED for Hugging Face**
- `frame_id`: Unique identifier for each frame
- `timestamp`: Time in seconds when frame was captured
- `frequency`: Audio frequency at capture time (Hz)
- `time_scale`: Playback speed multiplier
- `capture_date`: ISO date string of capture

## Intended Use

This dataset is intended for training machine learning models on audio visualization patterns, waveform classification, or generative AI tasks.