菜单
Generates spoken audio with the ElevenLabs text-to-speech API. Use when converting text, summaries, or narration into an audio file or when the user asks to play generated speech aloud.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/akornmeier/skillz --skill elevenlabs-tts命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Design-engineering guidance for UI motion, interaction polish, gesture physics, Apple-style fluid interfaces, animation terminology, motion opportunity audits and plans, animation code reviews, and throwaway interaction prototypes. Use when animation or transition behavior is being designed, built, reviewed, debugged, named, or evaluated, or when the user asks what should animate or requests an interactive UI prototype. Do not use for static visual styling, general UX research, unrelated code prototypes, or D3 implementation unless explicitly invoked.
Monitors GitHub pull requests through CI and automated review. Use when the user asks to watch, monitor, or babysit a PR, investigate failed checks, or respond to review-bot comments.
Automates browsers with Playwright CLI for UI testing, screenshots, navigation, scraping, and parallel browser sessions. Use when a task requires headless browser interaction or verification without a user-driven browser.
正在显示 SKILL.md
| name | elevenlabs-tts |
| description | Generates spoken audio with the ElevenLabs text-to-speech API. Use when converting text, summaries, or narration into an audio file or when the user asks to play generated speech aloud. |
| compatibility | Requires Python 3, curl 7.76 or newer, network access, and ELEVENLABS_API_KEY. Generation may incur paid ElevenLabs usage and requires explicit approval. Playback depends on the host operating system and installed audio tools. |
| metadata | {"category":"audio-generation"} |
Generate an MP3 only after the user approves potentially paid ElevenLabs usage for the current task. Never expose the API key or send unapproved text to the service.
Confirm the exact text, output path, voice ID, and whether playback is requested.
Confirm explicit approval for external, potentially paid generation.
Preflight without making a request:
python3 --version
curl --version
test -n "$ELEVENLABS_API_KEY"
Stop with a clear explanation if a dependency or key is unavailable. Do not install software or create credentials.
Put nontrivial narration in a UTF-8 text file, then run the bundled generator using an absolute skill path:
python3 <skill-directory>/scripts/elevenlabs_tts.py \
--text-file narration.txt \
--output narration.mp3 \
--voice-id <voice-id> \
--model-id eleven_multilingual_v2 \
--approve-paid-use
Use --force only after confirming that overwriting the target is intended. The script does not retry POST requests because a retry could create a duplicate charge.
Verify the script reports a successful audio response and a non-empty MP3 written atomically to the requested path.
If playback was requested, run:
python3 <skill-directory>/scripts/play_audio.py narration.mp3
Playback failure does not invalidate a successfully generated file. Report the file path and playback limitation separately.
Read references/elevenlabs-options.md when selecting a voice, model, output behavior, or local mock endpoint. Treat example IDs as changeable service identifiers, not permanent guarantees.