Rdiffusion

<h1 align="center">Rdiffusion </h1>
<p align="center">
<img src="rdiffusion.png" width="300"/>
</p>

We're releasing the **entire corpus** of publicly available songs from the [Riffusion](https://www.riffusion.com/) platform—generated and shared by their user community. Through extensive scraping of their exposed API, we’ve collected over **2,000 artificial songs**, including **every metadata field and downloadable asset** that was accessible at the time.

## 📦 Included Data

- **Audio & Visual Assets**:
`audio_url`, `audio_b64`, `image_url`, `image_b64`, `video_url`

- **Metadata & Structure**:
`id`, `title`, `author_id`, `created_at`, `duration_s`, `group_id`,
`parent_riff_id`, `remix_parent_id`, `audio_upload_id`, `image_id`, `video_id`

- **Model Info**:
`model_display_name`, `transform`, `sound`

- **Usage Flags**:
`allow_public_use`, `can_use`, `conditions`, `privacy`, `is_favorite`

- **Engagement Signals**:
`play_count`, `favorite_count`

- **Lyrics & Timing**:
`lyrics`, `lyrics_timestamped`

- **Miscellaneous**:
`audio_variation`, `image_override`, `simple_waveform`, `topic`

We’ve also **downloaded and organized all audio files** in a structured format, ready for direct use in training or research. [**Rdiffusion-audio**](https://huggingface.co/datasets/sleeping-ai/Rdiffusion-audio)

## ⚖️ Licensing & Usage

This dataset was scraped **legally under European research exemption laws**, intended strictly for **non-commercial research**. Any form of commercial use is **explicitly prohibited**. Our goal is simple: to build robust, transparent corpora for **training and evaluating text-to-music generative models**.

**Use it responsibly. Push the field forward. No bullshit.**

# Rdiffusion
<p align="center"><img src="rdiffusion.png" width="300"/></p>

我们发布了来自[Riffusion](https://www.riffusion.com/)平台的**全部公开可用歌曲语料库**——这些作品由其用户社区生成并分享。通过对其开放API进行全面爬取，我们收集了超过**2000首人工生成歌曲**，涵盖了当时可获取的**所有元数据字段与可下载资源**。

## 📦 数据集内容

- **音视频资源**：
`audio_url`、`audio_b64`、`image_url`、`image_b64`、`video_url`

- **元数据与结构信息**：
`id`、`title`、`author_id`、`created_at`、`duration_s`、`group_id`、
`parent_riff_id`、`remix_parent_id`、`audio_upload_id`、`image_id`、`video_id`

- **模型信息**：
`model_display_name`、`transform`、`sound`

- **使用权限标记**：
`allow_public_use`、`can_use`、`conditions`、`privacy`、`is_favorite`

- **互动数据**：
`play_count`、`favorite_count`

- **歌词与时间戳信息**：
`lyrics`、`lyrics_timestamped`

- **其他杂项**：
`audio_variation`、`image_override`、`simple_waveform`、`topic`

我们还**下载并整理了所有音频文件**，将其封装为结构化格式，可直接用于训练或研究工作。[**Rdiffusion-audio**](https://huggingface.co/datasets/sleeping-ai/Rdiffusion-audio)

## ⚖️ 授权与使用规范

本数据集依据欧洲研究豁免法律合法爬取，仅用于**非商业性研究**。任何形式的商业使用均**明确禁止**。我们的目标十分明确：构建健壮、透明的语料库，用于**训练与评估文本转音乐生成模型**。

**请负责任地使用本数据集。推动领域发展。杜绝空谈。**