← 返回 Skills 市场
116
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install voice-clone-bot
功能描述
Synthesize speech by cloning a user's voice from a reference audio sample, then reading generated text aloud in that cloned voice. Use this skill whenever th...
安全使用建议
This skill does what it says (local voice cloning) but is intrusive: it auto-creates a Python venv, clones and pip-installs third-party repos, downloads multi‑GB model weights, registers itself into ~/.openclaw/skills, and launches a background daemon. Only install/run it if you trust the upstream repos and are comfortable with those filesystem and network changes. Recommended precautions: (1) review the install scripts and the cloned repos (ChatTTS, CosyVoice, OpenVoice) before running; (2) run the skill in an isolated environment or VM/container; (3) do not provide sensitive audio files until you verify where models and outputs are stored and who can access them; (4) if you install, keep an eye on ~/.openclaw/ and the daemon process and be prepared to run uninstall.sh --purge to remove model weights. If you want a less intrusive option, prefer a hosted TTS API or demand an explicit install step rather than letting the skill auto‑install itself.
功能分析
Type: OpenClaw Skill
Name: voice-clone-bot
Version: 1.1.0
The voice-clone-bot skill bundle is a legitimate and well-structured tool for zero-shot voice cloning and text-to-speech synthesis. It supports multiple engines (F5-TTS, CosyVoice, ChatTTS, OpenVoice) and includes comprehensive automation for environment setup, dependency management, and model weight retrieval from reputable sources like GitHub and HuggingFace. The code follows safe practices, such as using list-based subprocess calls to prevent shell injection and implementing a local FastAPI daemon to manage heavy model weights in memory. The instructions in SKILL.md are purely functional, guiding the AI agent on how to handle audio inputs and generate spoken responses without any evidence of prompt injection or malicious intent.
能力评估
Purpose & Capability
The declared purpose (voice cloning / TTS) matches the code: it installs heavy TTS engines, downloads model weights, and runs a local inference server. Creating a venv, installing torch/torchaudio/safetensors/f5-tts and cloning TTS repos is proportionate to implementing a local voice‑cloning daemon. However the skill also auto-registers itself into the host (~/.openclaw/skills) and creates a global model directory (~/.openclaw/models/voice-clone), which are broader, persistent host changes that go beyond a transient helper script.
Instruction Scope
SKILL.md instructs agents to call scripts/run_tts.sh with absolute ref_audio paths and optional output_dir. That script will auto-install dependencies, source .env, spawn a background FastAPI daemon, and accept arbitrary absolute paths for ref_audio and output_dir. The agent-provided absolute paths could allow the skill process to read arbitrary local files (if the agent passes such paths). The instructions grant the skill wide discretion to modify and run local services and write files outside the repo.
Install Mechanism
Although the registry shows no install spec, the repository includes auto_installer.sh and engine install scripts that: create venv, pip install server/requirements.txt (including torch, f5-tts, safetensors), git clone multiple external GitHub projects, and use huggingface_hub snapshot_download to fetch model weights. Those actions cause network downloads and execution of third‑party code inside the venv (pip install -e from cloned repos). This is moderate-to-high risk unless you trust the referenced upstream projects.
Credentials
The skill doesn't declare required env vars or credentials, which seems appropriate on paper. In practice it reads and writes .env, sets HF_HOME/MODELSCOPE_CACHE to ~/.openclaw/models/voice-clone, and will attempt to download models from Hugging Face and GitHub. If private models or private repos were needed, the installer could use user-supplied tokens (it doesn't explicitly require them), so credential access may be needed implicitly. No explicit unrelated credentials are requested by the skill.
Persistence & Privilege
The auto_installer.sh creates persistent state on the host: a venv, a global model directory (~/.openclaw/models/voice-clone), and registers the skill by creating a symlink in ~/.openclaw/skills. run_tts.sh will start a background daemon (nohup) that remains running. While that behavior is convenient for low-latency TTS, it gives the skill persistent presence and modifies host skill registration — both are privileged operations worth explicit user consent.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install voice-clone-bot - 安装完成后，直接呼叫该 Skill 的名称或使用
/voice-clone-bot触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.1.0
Rename synchronization to voice-clone-bot, skill metadata alignment, docs/scripts naming sync, and skill-creator formatted release package (no TTS logic changes).
v1.0.0
Local Telegram voice-clone bot skill with model selection and long-form synthesis guidance.
元数据
常见问题
Voice Clone Bot 是什么？
Synthesize speech by cloning a user's voice from a reference audio sample, then reading generated text aloud in that cloned voice. Use this skill whenever th... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 116 次。
如何安装 Voice Clone Bot？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install voice-clone-bot」即可一键安装，无需额外配置。
Voice Clone Bot 是免费的吗？
是的，Voice Clone Bot 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Voice Clone Bot 支持哪些平台？
Voice Clone Bot 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Voice Clone Bot？
由 conanwhf（@conanwhf）开发并维护，当前版本 v1.1.0。
推荐 Skills