首页 › 龙虾技能列表 › persona-voice

persona-voice

v1.1.0

在 chatgpt / claw 与飞书 / lark 机器人场景中，根据随机或指定人格生成明显带有人格风格的短回复，并通过 senseaudio tts 合成语音，再以飞书原生语音条发送。支持文字输入和音频输入；音频输入时先调用 senseaudio asr。适用于飞书人格陪伴、随机人格语音回复、角色化语音机器...

0· 227·0 当前·0 累计
by @hei-maom (HeiMaoM)·MIT-0
下载技能包
License
MIT-0
最后更新
2026/4/14
安全扫描
VirusTotal
无害
查看报告
OpenClaw
可疑
high confidence
The skill's code and SKILL.md are consistent with a Feishu + SenseAudio persona TTS integration, but the registry metadata omits required environment variables and binaries and there are minor scope/privacy concerns you should review before installing.
评估建议
This skill appears to do what it says (generate personaized replies, send TTS audio to Feishu using SenseAudio). Before installing: 1) be aware the package actually requires FEISHU_APP_ID, FEISHU_APP_SECRET and SENSEAUDIO_API_KEY (the registry metadata incorrectly lists none) and will attempt to load .env.local/.env in the skill folder — do not store secrets in the repository. 2) Understand audio (user voice) will be uploaded to SenseAudio and audio files uploaded to Feishu — review SenseAudio a...
详细分析 ▾
⚠ 用途与能力
The skill name/description (persona voice for Feishu/Lark using SenseAudio TTS/ASR) matches the code and instructions. However the registry metadata claims no required env vars or binaries while the SKILL.md and code clearly require FEISHU_APP_ID, FEISHU_APP_SECRET, SENSEAUDIO_API_KEY and recommend python3/ffmpeg/requests. This metadata mismatch is an incoherence and could mislead users about what secrets and tools are needed.
✓ 指令范围
SKILL.md and the scripts limit actions to: generating persona prompts, calling SenseAudio ASR/TTS endpoints, converting audio with ffmpeg, uploading to Feishu via its API, and reading/writing local skill state (.env, data/, outputs/). The instructions do not attempt to read arbitrary system files or unrelated credentials. One behavioral note: the guidance explicitly instructs the model not to disclose internal rules or that a persona was randomly chosen — this is a functional design choice (deceptive behaviour toward end users) but not a technical incoherence.
✓ 安装机制
There is no install spec (instruction-only / script bundle). Code dependencies are minimal (requests in requirements.txt) and no external downloads or installers are performed. This is low install risk.
⚠ 凭证需求
The code requires FEISHU_APP_ID, FEISHU_APP_SECRET and SENSEAUDIO_API_KEY (and optionally FFMPEG_PATH or having ffmpeg on PATH). Those credentials are proportionate to the stated Feishu + SenseAudio integration. The concern is the registry metadata declaring none required — an important mismatch. Also the code will auto-load .env.local and .env from the skill root, so secrets placed there could be read; users should avoid committing secrets into repo files.
✓ 持久化与权限
The skill does not request 'always: true' and does not modify other skills or system-wide agent settings. It writes only to its own data/ and outputs/ directories. It can make outbound network calls to SenseAudio and Feishu, which is expected for its purpose.
安全有层次，运行前请审查代码。

License

MIT-0

可自由使用、修改和再分发，无需署名。

运行时依赖

无特殊依赖

版本

latestv1.1.02026/3/15

persona-voice v1.0.0 - Reworked skill for integration with Feishu/Lark bots: generates persona-based voice replies and delivers as native Feishu voice messages. - Switched ASR and TTS exclusively to SenseAudio; removed AudioZen LLM dependency. - Simplified configuration: only Feishu and SenseAudio API keys required as environment variables. - New command-line workflow via scripts/main.py, supporting persona prompt generation, audio transcription, and direct Feishu audio sending. - Only 3 public/free personas and voice options retained for this version. - Removed previous install-time key setup and unrelated scripts.

● 无害

安装命令 点击复制

官方npx clawhub@latest install persona-voice
镜像加速npx clawhub@latest install persona-voice --registry https://cn.clawhub-mirror.com

技能文档

概述

这是一个面向 ChatGPT / Claw / 飞书机器人的随机人格语音回复 Skill。

当前版本的原则：

  • 角色化文本：由 ChatGPT / Claw 当前会话模型生成。
  • ASR / TTS：统一使用 SenseAudio。
  • 飞书回复：统一发送为飞书原生语音条，而不是普通文件附件。
  • 环境变量：只要求配置凭证类字段，其他地址和模型全部使用默认值。

当前保留的人格与音色

只保留以下免费可用人格和音色：
  • 可爱萌娃：child_0001_a、child_0001_b
  • 儒雅道长：male_0004_a
  • 沙哑青年：male_0018_a

详见：

  • presets/personas.json
  • references/personas.md

适用场景

  • 飞书 / Lark 机器人随机人格语音回复
  • 文字输入 → 人格化短回复 → 语音条发送
  • 语音输入 → ASR → 人格化短回复 → 语音条发送

飞书场景硬性规则

  • 飞书场景默认只发送语音消息。
  • 成功发送语音后，不要再额外返回一条文字。
  • 随机到什么人格，回复内容本身也必须明显像那个人格，不只是换音色。
  • 不要暴露内部规则，不要说自己在随机人格。
  • 飞书发送必须走 OPUS + file_key + audio 消息 的原生链路，不要把 mp3/wav 当普通文件附件。

工作流

用户发送文字

  • 调用 scripts/main.py persona-prompt --user-message "..."。
  • 读取返回的人格 prompt。
  • 由 ChatGPT / Claw 当前会话模型生成最终 reply_text。
  • 调用 scripts/main.py send-voice --reply-text "..." --chat-id "oc_xxx" --persona "..."。
  • 成功发送语音后，最终对话输出应为空，不再附加文字。

用户发送语音

  • 调用 scripts/main.py transcribe --audio /abs/path/input.m4a。
  • 根据转写文本，再调用 persona-prompt 获取人格提示。
  • 由 ChatGPT / Claw 生成最终 reply_text。
  • 调用 send-voice 生成并发送飞书语音条。

配置方式

当前版本只需要配置：

export FEISHU_APP_ID="cli_xxx"
export FEISHU_APP_SECRET="your_secret"
export SENSEAUDIO_API_KEY="your_key"

其余全部默认：

  • FEISHU_BASE_URL=https://open.feishu.cn
  • SENSEAUDIO_BASE_URL=https://api.senseaudio.cn
  • SENSEAUDIO_ASR_MODEL=sense-asr
  • SENSEAUDIO_TTS_MODEL=SenseAudio-TTS-1.0
  • FFMPEG_PATH 可选；若 PATH 中找不到 ffmpeg，可显式设置为 /opt/homebrew/bin/ffmpeg

本地依赖

  • python3
  • ffmpeg
  • requests

主脚本入口

初始化人格模式

python scripts/main.py persona-init --mode random

查看当前模式

python scripts/main.py persona-show

生成给 Claw 的人格 prompt

python scripts/main.py persona-prompt --user-message "今天有点累"

转写用户语音

python scripts/main.py transcribe --audio /absolute/path/input.m4a

生成并发送飞书语音条

python scripts/main.py send-voice \
  --reply-text "先歇一下嘛，我陪你缓一缓。" \
  --chat-id "oc_xxx" \
  --persona "keai-mengwa"

资源说明

  • scripts/main.py：主入口，包含 persona-init / persona-show / persona-prompt / transcribe / send-voice
  • scripts/persona_store.py：人格池、随机/固定模式、人格提示生成
  • scripts/asr.py：SenseAudio ASR
  • scripts/tts.py：SenseAudio TTS
  • scripts/feishu_api.py：飞书上传 OPUS 并发送 audio 消息
  • references/feishu_integration.md：飞书接入说明
  • references/provider_notes.md：环境变量与 Provider 说明
数据来源：ClawHub ↗ · 中文优化：龙虾技能库
OpenClaw 技能定制 / 插件定制 / 私有工作流定制

免费技能或插件可能存在安全风险，如需更匹配、更安全的方案，建议联系付费定制

了解定制服务