← 返回 Skills 市场
Speech To Text
作者
Ömer Karışman
· GitHub ↗
· v0.1.5
3704
总下载
0
收藏
35
当前安装
2
版本数
在 OpenClaw 中安装
/install speech-to-text
功能描述
Transcribe audio to text with Whisper models via inference.sh CLI. Models: Fast Whisper Large V3, Whisper V3 Large. Capabilities: transcription, translation,...
安全使用建议
Install the inference.sh CLI only if you trust the provider, and prefer the documented manual checksum verification path when possible. Review `infsh` commands before running them, and do not submit confidential, regulated, or private recordings unless inference.sh is approved for that data under your account.
功能分析
Type: OpenClaw Skill
Name: speech-to-text
Version: 0.1.5
The skill bundle is classified as suspicious primarily due to the installation method described in `SKILL.md`. It instructs the AI agent to execute `curl -fsSL https://cli.inference.sh | sh`, which downloads and runs a shell script directly from a remote server. This practice is a significant security risk, creating a supply chain vulnerability and potential for Remote Code Execution (RCE) if the `cli.inference.sh` domain or server were compromised. While the documentation attempts to explain the script's benign nature, the method itself is inherently insecure and could lead to arbitrary code execution on the host system.
能力评估
Purpose & Capability
The name, description, examples, and model list consistently describe transcribing, translating, timestamping, and captioning audio through inference.sh Whisper apps.
Instruction Scope
Runtime tool scope is limited to `Bash(infsh *)`, which is broader than only transcription commands but still confined to the inference.sh CLI and matches the documented workflows.
Install Mechanism
The quick start recommends `curl -fsSL https://cli.inference.sh | sh && infsh login`; the artifact discloses this remote installer and mentions checksum/manual verification, but users still need to trust the installer source.
Credentials
Submitting audio URLs to inference.sh is expected for a cloud transcription skill, but recordings may contain sensitive meeting, interview, or voice-note content and the privacy warning could be clearer.
Persistence & Privilege
The skill requires `infsh login`, implying a local service session, but the artifact shows no elevated permissions, background processes, destructive actions, unrelated credential access, or hidden persistence.
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 speech-to-text - 安装完成后，直接呼叫该 Skill 的名称或使用
/speech-to-text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
v0.1.5
- Updated documentation for clear setup instructions using inference.sh CLI.
- Detailed available Whisper model options, usage examples, and input formats.
- Added new sections on extracting audio from video, translation, and video subtitle workflows.
- Enhanced guidance for supported languages and output structure.
- Improved 'Related Skills' for easy access to complementary AI tools.
v0.1.0
- Initial release of speech-to-text skill.
- Transcribe audio to text using Whisper models via inference.sh CLI.
- Supports transcription, translation, multi-language, and timestamps.
- Includes Fast Whisper Large V3 and Whisper V3 Large model options.
- Provides example workflows for meetings, podcasts, subtitles, and more.
- Output is returned as structured JSON with text, segments, and detected language.
元数据
常见问题
Speech To Text 是什么？
Transcribe audio to text with Whisper models via inference.sh CLI. Models: Fast Whisper Large V3, Whisper V3 Large. Capabilities: transcription, translation,... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 3704 次。
如何安装 Speech To Text？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install speech-to-text」即可一键安装，无需额外配置。
Speech To Text 是免费的吗？
是的，Speech To Text 完全免费（开源免费），可自由下载、安装和使用。
Speech To Text 支持哪些平台？
Speech To Text 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 Speech To Text？
由 Ömer Karışman（@okaris）开发并维护，当前版本 v0.1.5。
推荐 Skills