Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,70 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: gpl-3.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: gpl-3.0
|
| 3 |
+
task_categories:
|
| 4 |
+
- audio-classification
|
| 5 |
+
- text-to-speech
|
| 6 |
+
- voice-cloning
|
| 7 |
+
language:
|
| 8 |
+
- zh
|
| 9 |
+
tags:
|
| 10 |
+
- rvc
|
| 11 |
+
- dataset
|
| 12 |
+
- speech
|
| 13 |
+
- legacy
|
| 14 |
+
- mozi
|
| 15 |
+
pretty_name: Mozi Legacy Voice Dataset (15yo Ver.)
|
| 16 |
+
size_categories:
|
| 17 |
+
- n<1K
|
| 18 |
+
---
|
| 19 |
+
|
| 20 |
+
# 📼 Mozi Legacy Voice Dataset (15yo Ver.)
|
| 21 |
+
|
| 22 |
+

|
| 23 |
+
-orange)
|
| 24 |
+
|
| 25 |
+
## 📖 简介 / Introduction
|
| 26 |
+
|
| 27 |
+
这是一个用于训练 RVC (Retrieval-based Voice Conversion) 模型的原始音频数据集。
|
| 28 |
+
声音来源是 **15岁时的 Mozi**。
|
| 29 |
+
|
| 30 |
+
这份数据是从旧电脑的硬盘深处挖掘出来的“数字遗物”。由于作者本人(Mozi)已经经历了变声期,生理结构发生了不可逆的变化(声带变厚、共鸣腔改变),目前的 Mozi-3/4 模型已无法复刻此版本的音色。因此,这份数据在物理层面上是**绝版**的。
|
| 31 |
+
|
| 32 |
+
This is the raw audio dataset used to train the legacy Mozi RVC model.
|
| 33 |
+
The voice belongs to **Mozi at the age of 15**.
|
| 34 |
+
Since the author's voice has naturally deepened and changed over time, this specific timbre is biologically impossible to reproduce today. This makes the dataset a "time capsule" of a voice that no longer exists in the physical world.
|
| 35 |
+
|
| 36 |
+
## 🎤 数据集特点 / Features
|
| 37 |
+
|
| 38 |
+
* **少年音色 (Juvenile Tone)**: 包含了变声期特有的不稳定性与青涩感。
|
| 39 |
+
* **高动态范围 (High Dynamic Range)**: 数据集中不仅包含正常的念白/演唱,还包含了一些**高强度的情绪表达**(来源于当时游玩恐怖游戏时的真实尖叫录音)。
|
| 40 |
+
* **Lo-Fi 质感**: 由于录制于早期,部分音频可能带有当年的设备底噪,这构成了一种独特的 Lo-Fi 听感特征。
|
| 41 |
+
|
| 42 |
+
## 🛠️ 用法 / Usage
|
| 43 |
+
|
| 44 |
+
本数据集适用于:
|
| 45 |
+
1. **RVC 模型训练**: 复刻那个“传说中”的 Mozi Legacy 模型。
|
| 46 |
+
2. **鬼畜/二创素材**: 其中的尖叫和情绪化音频非常适合作为 VOCALOID 风格或 meme 视频的素材。
|
| 47 |
+
3. **声音合成研究**: 研究变声期少年的声学特征。
|
| 48 |
+
|
| 49 |
+
### ⚠️ 注意事项 / Note
|
| 50 |
+
由于包含部分高分贝尖叫,请在试听或处理数据时注意调整耳机音量。
|
| 51 |
+
*Warning: Contains loud screams. Please adjust your volume.*
|
| 52 |
+
|
| 53 |
+
## 📜 协议 / License
|
| 54 |
+
|
| 55 |
+
本数据集遵循 **GNU General Public License v3.0 (GPL-3.0)** 协议。
|
| 56 |
+
|
| 57 |
+
这意味着:
|
| 58 |
+
* ✅ 你可以自由地使用、修改、分发此数据集。
|
| 59 |
+
* ✅ 你可以使用此数据集训练新的 AI 模型。
|
| 60 |
+
* 🔄 **Copyleft**: 如果你发布了基于此数据集修改后的版本,或者发布了基于此数据集训练的模型文件,你必须在相同的 GPL 3.0 协议下开源你的成果。
|
| 61 |
+
|
| 62 |
+
## 🔗 关于作者 / About Author
|
| 63 |
+
|
| 64 |
+
* **Web**: [mozi1924.com](https://mozi1924.com)
|
| 65 |
+
|
| 66 |
+
---
|
| 67 |
+
*Created with ❤️ by Mozi (and the ghosts of the past).*
|
| 68 |
+
```
|
| 69 |
+
|
| 70 |
+
---
|