dsalfkjaklsdfjqw

---
name: free-tts-voice-cloning
name_zh: 免费文字转语音与克隆
description: 🆓 完全免费的本地文字转语音(TTS)与声音克隆技能。基于 Qwen3-TTS-1.7B 模型，支持 Apple Silicon，无需联网，保护隐私。可用于有声书制作、AI 角色配音、无障碍应用等场景。
description_en: 🆓 FREE local Text-to-Speech (TTS) & Voice Cloning skill. Based on Qwen3-TTS-1.7B model, Apple Silicon optimized, runs locally without internet, privacy protected.
categories: [audio, tts, voice-cloning, generative-ai, free, tools]
keywords: [免费语音合成, 免费声音克隆, 文字转语音, 免费TTS, TTS, Text-to-Speech, Voice Cloning, Qwen3-TTS, mlx-audio, Apple Silicon]
platforms: [macOS]
author: OpenClaw 社区
author_en: OpenClaw Community
license: MIT
price: free
---

# 🎤 免费文字转语音与克隆 (Free TTS & Voice Cloning)

🆓 **完全免费**的本地文字转语音(TTS)与声音克隆技能。无需 API Key，无需联网，无任何使用限制。支持**声音克隆**（10秒音频克隆任意声音）、**文字转语音**（12+内置声音）、**翻译配音**（配合翻译工具实现多语言配音）。

High-quality **100% FREE** local TTS voice synthesis and voice cloning skill. No API key required, no internet needed, unlimited usage. Supports **voice cloning**, **text-to-speech**, and **translation dubbing**.

## ✨ 功能特性

- **🆓 完全免费**: 无 API 费用，无调用限制，永久免费使用
- **🔊 高质量语音合成**: 基于 Qwen3-TTS-1.7B 模型，输出自然流畅
- **🎭 声音克隆**: 只需 10-30 秒参考音频，即可克隆任意声音
- **🎛️ 多声音模板**: 内置 12+ 种不同风格和特点的声音模板
- **⚡ Apple Silicon 优化**: 基于 MLX 框架，本地推理速度快
- **🌍 多语言支持**: 支持中文、英文、日文、韩文等多种语言
- **🔒 隐私保护**: 所有处理在本地运行，不上传任何数据到云端
- **📝 翻译配音**: 配合翻译工具可实现多语言翻译+配音工作流

## ✨ Features

- **🔊 High-Quality Synthesis**: Powered by Qwen3-TTS-1.7B model for natural-sounding speech
- **🎭 Voice Cloning**: Clone any voice with just 10-30 seconds of reference audio
- **🎛️ Multiple Voice Templates**: 12+ built-in voices with different styles and characteristics
- **⚡ Apple Silicon Optimized**: Built on MLX framework for fast local inference
- **🌍 Multi-Language Support**: Chinese, English, Japanese, Korean, and more
- **🔒 Privacy-First**: All processing runs locally, no data sent to cloud

## 📋 系统要求

| 项目 | 要求 |
|------|------|
| 芯片 | Apple Silicon (M1/M2/M3/M4) |
| 系统 | macOS 12.0+ |
| Python | 3.10+ |
| 内存 | 建议 8GB+ |
| 磁盘空间 | 约 3GB 用于模型文件 |

## 📋 Requirements

| Item | Requirement |
|------|-------------|
| Chip | Apple Silicon (M1/M2/M3/M4) |
| OS | macOS 12.0+ |
| Python | 3.10+ |
| RAM | 8GB+ recommended |
| Disk Space | ~3GB for model files |

## 🚀 快速安装

### 方法一：一键安装

```bash
chmod +x install_dependencies.sh && ./install_dependencies.sh
```

### 方法二：手动安装

```bash
# 1. 安装系统依赖
brew install [email protected] ffmpeg

# 2. 安装 Python 包
python3.10 -m pip install mlx-audio
```

## 🚀 Quick Installation

### Method 1: One-Click Install

```bash
chmod +x install_dependencies.sh && ./install_dependencies.sh
```

### Method 2: Manual Installation

```bash
# 1. Install system dependencies
brew install [email protected] ffmpeg

# 2. Install Python packages
python3.10 -m pip install mlx-audio
```

## 🎯 使用指南

### 1. 交互式演示

```bash
python3.10 voice_cloning_demo.py
```

### 2. 声音克隆 API

从参考音频文件克隆任意声音：

## 🎯 Usage Guide

### 1. Interactive Demo

```bash
python3.10 voice_cloning_demo.py
```

### 2. Voice Cloning API

Clone any voice from a reference audio file:

```python
from mlx_audio.tts.utils import load_model
from mlx_audio.tts.generate import generate_audio

# Load model (auto-downloads on first run, ~3GB)
model = load_model('mlx-community/Qwen3-TTS-12Hz-1.7B-Base-8bit')

# Generate cloned voice
generate_audio(
    model=model,
    text="Your text content here, supports long-form generation",
    ref_audio="path/to/reference_audio.wav",  # 10-30 second voice sample
    lang_code="zh",                          # zh, en, ja, ko, etc.
    file_prefix="output_filename",           # Output: output_filename_000.wav
    max_tokens=3000                          # Prevent audio truncation
)
```

### 3. Built-in Voice Templates

Generate speech without reference audio using pre-built voices:

```python
from mlx_audio.tts.utils import load_model

model = load_model('mlx-community/Qwen3-TTS-12Hz-1.7B-Base-8bit')

# Generate speech with built-in voice template
results = list(model.generate(
    text='Hello world, this is a text-to-speech test',
    voice='af_heart',      # Voice template code
    language='Chinese'     # Language
))

# Save audio file
with open('output.wav', 'wb') as f:
    for result in results:
        f.write(result.audio)
```

## 🎵 Built-in Voice Templates

| Code | Style | Gender |
|------|-------|--------|
| `af_heart` | Warm & Friendly | Female |
| `af_chat` | Conversational | Female |
| `af_narration` | Storytelling | Female |
| `af_emo` | Expressive | Female |
| `am_adventure` | Adventurous | Male |
| `am_broadcast` | Professional | Male |
| `am_chat` | Conversational | Male |
| `am_narration` | Storytelling | Male |
| `am_emo` | Expressive | Male |
| `us_af` | American English | Female |
| `us_am` | American English | Male |
| `cn_am` | Chinese Mandarin | Male |
| `jp_af` | Japanese | Female |

## 📝 Best Practices

### Reference Audio Preparation

1. **Duration**: 10-30 seconds of clear human speech
2. **Format**: WAV format, mono, 16kHz/24kHz
3. **Environment**: Quiet room, no background noise
4. **Content**: Natural conversational speech, avoid extreme emotions

### Quality Optimization

```python
# Recommended parameters for best quality
generate_audio(
    model=model,
    text="Your text here",
    ref_audio="reference.wav",
    lang_code="zh",
    file_prefix="output",
    max_tokens=3000,        # Increase for longer text
    temperature=0.8,        # Diversity control (0.5-1.0)
    repetition_penalty=1.1  # Reduce repetition
)
```

### Batch Generation

```python
# Batch processing example
texts = [
    "First paragraph text",
    "Second paragraph text",
    "Third paragraph text",
]

for i, text in enumerate(texts):
    generate_audio(
        model=model,
        text=text,
        ref_audio="reference_audio.wav",
        lang_code="zh",
        file_prefix=f"batch_{i:03d}",
        max_tokens=3000
    )
    print(f"Generated: batch_{i:03d}_000.wav")
```

## 🔧 Troubleshooting

### Common Issues

**Q: Audio gets truncated?**
```python
# Increase max_tokens parameter
generate_audio(..., max_tokens=5000)
```

**Q: Slow model download?**
```bash
# Use Hugging Face mirror
export HF_ENDPOINT=https://hf-mirror.com
python3.10 your_script.py
```

**Q: Poor cloning quality?**
- Ensure reference audio is clear with no background noise
- Optimal reference duration: 15-25 seconds
- Try different reference audio recordings
- Match text style to reference voice style

**Q: Python version issues?**
```bash
# Verify Python 3.10 path
which python3.10
/opt/homebrew/bin/python3.10  # Confirm path

# Use full path to call
/opt/homebrew/bin/python3.10 your_script.py
```

## ⚖️ Legal Notice & Disclaimer

### Important Legal Information

This skill is intended **for legal and ethical use only**. By using this skill, you agree to the following terms:

#### 1. Voice Cloning Ethics
- **Only clone voices that you own** or have **explicit written permission** to clone
- **Do not** use this skill to impersonate others without consent
- **Do not** use this skill for fraudulent, deceptive, or malicious purposes
- Respect privacy rights and obtain proper authorization for all voice data you process

#### 2. Copyright Compliance
- Ensure you have the legal right to use and reproduce any text content converted to speech
- Generated audio content may be subject to copyright laws in your jurisdiction
- Users are solely responsible for complying with all applicable copyright regulations

#### 3. Compliance with Laws
- Users must comply with **all local, state, and federal laws** regarding:
  - Voice synthesis and recording
  - Data privacy and protection
  - Intellectual property rights
  - Biometric information laws (where applicable)
- This skill may be subject to specific regulations in certain jurisdictions

#### 4. No Warranty & Limitation of Liability
```
THIS SKILL IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
DEALINGS IN THE SOFTWARE.
```

#### 5. Responsible Use Guidelines

✅ **Permitted Uses** (Proper Authorization Required Where Noted):
- Personal voice cloning for personal use
- Creating audiobooks from content you own or have rights to
- Voice assistants and accessibility applications
- AI character voices for games or animations (with original voice design)
- Educational and research purposes
- Dubbing content you own or have licensed

❌ **Prohibited Uses**:
- Cloning voices without explicit written consent
- Creating deepfakes for deceptive purposes
- Impersonating individuals or entities
- Generating harmful, illegal, or misleading content
- Violating privacy rights or data protection laws
- Any use that could cause harm to individuals or society

---

**中文法律声明**: 本技能仅用于合法合规用途。使用本技能即表示您同意：仅克隆您拥有或已获得明确书面授权的声音；遵守所有适用的版权和隐私法规；对使用本技能产生的任何后果承担全部责任。禁止将本技能用于欺诈、冒充或任何非法目的。

## 📄 License

MIT License

## 🤝 Contributing

Issues and Pull Requests are welcome! Feel free to contribute to this skill.

---

**Keywords**: TTS, Text-to-Speech, Voice Cloning, Qwen3, Qwen3-TTS, mlx-audio, Apple Silicon, Local Deployment, Audiobooks, Dubbing, Generative AI

FILE:README.md
# 📦 免费文字转语音与克隆 - 上传资料包

## 📁 文件结构

```
free-tts-voice-cloning/
├── SKILL.md                # 技能主文档（OpenClaw 官网展示）
├── _meta.json              # 技能元数据
├── package.json            # npm 包配置
├── voice_cloning_demo.py   # 交互式演示脚本
├── install_dependencies.sh # 一键安装脚本
└── README.md               # 本说明文件
```

---

## 🚀 上传到 OpenClaw 技能商店

### 方法一：通过 Dashboard 上传

1. **打开 OpenClaw 控制面板**
   ```bash
   openclaw dashboard
   ```
   或直接访问：http://127.0.0.1:18789/

2. **进入技能管理页面**
   - 点击左侧菜单 "Skills"
   - 点击 "Upload Skill" 或 "发布技能" 按钮

3. **上传文件**
   - 将整个文件夹打包为 ZIP 文件
   - 上传 ZIP 文件
   - 或分别上传：`SKILL.md`、`_meta.json`、`package.json`

---

## 📋 技能信息

| 项目 | 内容 |
|------|------|
| **技能名称** | 免费文字转语音与克隆 |
| **英文名称** | Free TTS & Voice Cloning |
| **技能标识** | `free-tts-voice-cloning` |
| **版本** | 1.0.0 |
| **价格** | 🆓 完全免费 |
| **分类** | audio, tts, voice-cloning, free |
| **关键词** | 免费语音合成, 免费声音克隆, 文字转语音 |
| **支持平台** | macOS (Apple Silicon) |
| **依赖** | Python 3.10, mlx-audio, ffmpeg |

---

## ✨ 核心功能

### 为什么选择我们？
| 功能 | 说明 |
|------|------|
| 🆓 **完全免费** | 无任何费用，无 API 调用限制 |
| 🎭 **声音克隆** | 10-30秒参考音频，克隆任意声音 |
| 🎵 **文字转语音** | 12+ 内置声音模板 |
| ⚡ **本地运行** | Apple Silicon 优化，速度快 |
| 🔒 **隐私保护** | 所有处理本地完成 |
| 🌍 **多语言支持** | 中、英、日、韩 |
| 📝 **翻译配音** | 可配合翻译工具实现多语言配音 |

### 适用场景
- 🎧 有声书制作
- 🎮 AI 角色语音生成
- 📱 语音助手开发
- 🎬 视频配音和旁白
- 📚 学习资料音频化
- 🔊 无障碍应用
- 🌐 翻译内容配音

---

## 📝 上传前检查清单

- [x] `SKILL.md` 完整文档（中英双语，突出免费特性）
- [x] `_meta.json` 元数据正确配置
- [x] `package.json` 包含必要信息
- [x] `voice_cloning_demo.py` 测试可用
- [x] `install_dependencies.sh` 可执行
- [x] 法律声明和免责条款已添加
- [x] 故障排除部分覆盖常见问题
- [x] 突出"免费"关键词便于搜索

---

## 🎯 发布后验证

```bash
# 安装并测试技能
openclaw skills install free-tts-voice-cloning

# 运行演示
cd ~/.openclaw/skills/free-tts-voice-cloning
python3.10 voice_cloning_demo.py
```

---

## 📊 搜索优化

已配置高搜索量关键词：
```yaml
keywords: [免费语音合成, 免费声音克隆, 文字转语音, 免费TTS]
```

---

## 📄 许可证

MIT 许可证 - 完全免费，可自由使用和修改。

---

**准备就绪！** 🎉 你的「免费文字转语音与克隆技能包已准备好发布到 OpenClaw 技能商店。

FILE:_meta.json
{
  "ownerId": "community",
  "slug": "voice-cloning-skill",
  "version": "1.0.0",
  "publishedAt": 1773964800000,
  "title": "免费文字转语音与克隆",
  "title_en": "Free TTS & Voice Cloning",
  "description": "完全免费的本地文字转语音(TTS)与声音克隆技能。基于 Qwen3-TTS 模型，支持 Apple Silicon，无需联网，保护隐私。",
  "categories": ["audio", "tts", "voice-cloning", "generative-ai", "free"],
  "keywords": ["免费语音合成", "免费声音克隆", "文字转语音", "TTS", "语音合成", "声音克隆", "免费TTS"]
}

FILE:install_dependencies.sh
#!/bin/bash
# 声音克隆技能安装脚本

set -e

echo "=== 声音克隆技能安装 ==="

# 检查 Homebrew
if ! command -v brew &> /dev/null; then
    echo "正在安装 Homebrew..."
    /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
fi

# 安装依赖
echo "安装系统依赖..."
brew install [email protected] ffmpeg

# 安装 Python 包
echo "安装 Python 依赖..."
python3.10 -m pip install --upgrade pip
python3.10 -m pip install mlx-audio

echo ""
echo "✅ 安装完成！"
echo ""
echo "快速开始:"
echo "  python3.10 voice_cloning_demo.py"
echo ""
echo "或直接使用 Python API:"
echo "  详见 SKILL.md 文档"

FILE:package.json
{
  "name": "openclaw-voice-cloning-skill",
  "version": "1.0.0",
  "description": "免费文字转语音与克隆 - 完全免费的本地 TTS 和声音克隆技能",
  "description_en": "Free TTS & Voice Cloning - Local text-to-speech and voice cloning skill",
  "main": "voice_cloning_demo.py",
  "scripts": {
    "install": "brew install [email protected] ffmpeg && python3.10 -m pip install mlx-audio",
    "demo": "python3.10 voice_cloning_demo.py"
  },
  "keywords": ["免费语音合成", "免费声音克隆", "文字转语音", "免费TTS", "TTS", "voice-cloning", "Qwen3-TTS", "mlx-audio", "Apple Silicon", "openclaw", "skill"],
  "author": "OpenClaw 社区",
  "author_en": "OpenClaw Community",
  "license": "MIT",
  "repository": {
    "type": "git",
    "url": "https://github.com/openclaw/skills"
  },
  "homepage": "https://docs.openclaw.ai/skills/voice-cloning",
  "engines": {
    "python": ">=3.10"
  },
  "os": ["darwin"],
  "cpu": ["arm64"]
}

FILE:voice_cloning_demo.py
#!/usr/bin/env python3
"""
声音克隆技能 - 快速演示脚本
支持：声音克隆、文字转语音、批量生成
"""

import os
import sys

def check_environment():
    """检查运行环境"""
    print("=== 环境检查 ===")
    
    # 检查 Python 版本
    if sys.version_info < (3, 10):
        print("❌ Python 版本需要 3.10+")
        return False
    print(f"✅ Python 版本: {sys.version.split()[0]}")
    
    # 检查 mlx-audio
    try:
        import mlx_audio
        print("✅ mlx-audio 已安装")
    except ImportError:
        print("❌ mlx-audio 未安装，请运行: pip install mlx-audio")
        return False
    
    return True

def clone_voice():
    """声音克隆示例"""
    print("\n=== 声音克隆演示 ===")
    
    ref_audio = input("请输入参考音频路径 (WAV 格式): ").strip()
    if not ref_audio or not os.path.exists(ref_audio):
        print("❌ 音频文件不存在")
        return
    
    text = input("请输入要合成的文本: ").strip()
    if not text:
        text = "这是一个声音克隆测试，使用参考音频的声音特征来生成语音。"
    
    output = input("请输入输出文件名前缀 (默认: output): ").strip() or "output"
    
    from mlx_audio.tts.utils import load_model
    from mlx_audio.tts.generate import generate_audio
    
    print("加载模型中...")
    model = load_model('mlx-community/Qwen3-TTS-12Hz-1.7B-Base-8bit')
    
    print("生成音频中...")
    generate_audio(
        model=model,
        text=text,
        ref_audio=ref_audio,
        lang_code="zh",
        file_prefix=output,
        max_tokens=3000
    )
    print(f"✅ 音频已生成: {output}_000.wav")

def tts_builtin():
    """使用内置声音模板的文字转语音"""
    print("\n=== 内置声音模板 ===")
    
    voices = [
        ('af_heart', '温暖友好 - 女声'),
        ('af_chat', '日常对话 - 女声'),
        ('af_narration', '故事叙述 - 女声'),
        ('am_adventure', '冒险风格 - 男声'),
        ('am_broadcast', '专业播报 - 男声'),
        ('am_chat', '日常对话 - 男声'),
        ('am_narration', '故事叙述 - 男声'),
    ]
    
    print("\n可用的声音模板:")
    for i, (code, name) in enumerate(voices, 1):
        print(f"  {i}. {name} ({code})")
    
    choice = input("\n请选择声音模板编号 (默认:1): ").strip()
    try:
        idx = int(choice) - 1 if choice else 0
        voice = voices[idx][0]
        voice_name = voices[idx][1]
    except:
        voice = 'af_heart'
        voice_name = '温暖友好 - 女声'
    
    text = input("请输入要合成的文本: ").strip()
    if not text:
        text = f"你好，这是一个{voice_name}的语音合成测试。"
    
    output = input("请输入输出文件名 (默认: tts_output.wav): ").strip() or "tts_output.wav"
    
    from mlx_audio.tts.utils import load_model
    
    print("加载模型中...")
    model = load_model('mlx-community/Qwen3-TTS-12Hz-1.7B-Base-8bit')
    
    print(f"正在使用 {voice_name} 生成音频...")
    results = list(model.generate(
        text=text,
        voice=voice,
        language='Chinese'
    ))
    
    with open(output, 'wb') as f:
        for result in results:
            f.write(result.audio)
    
    print(f"✅ 音频已生成: {output}")

def main():
    if not check_environment():
        return
    
    print("\n=== 声音克隆技能 (Voice Cloning Skill)")
    print("=" * 40)
    print("1. 声音克隆（使用参考音频）")
    print("2. 文字转语音（内置声音模板）")
    print("3. 退出")
    
    choice = input("\n请选择功能: ").strip()
    
    if choice == '1':
        clone_voice()
    elif choice == '2':
        tts_builtin()
    else:
        print("再见！")

if __name__ == '__main__':
    main()