360 Secure SkillHub
360 安全技能中心

Clone Wizard

这是SenseAudio声音克隆引导工具,核心功能含:提供专业录音指导,对上传音频做噪声等质量检测并给建议,引导用户在平台完成克隆,再用voice_id生成AI语音试听。适用于用户想克隆个人声音、定制音色的场景,如问「怎么克隆声音」时触发。需搭配SenseAudio API密钥,输入为音频文件、voice_id,输出为质检报告、AI语音试听文件。

scikkk
scikkk
热度 51.16
最近更新 2026-05-11 17:32:01
分类:创作依赖:代码执行 / 第三方 API应用场景:API 集成 / 工作流 / 自动化脚本

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add clone-wizard
Metadata
name
senseaudio-clone-wizard
description
Guided voice cloning workflow — from recording tips to first playback. Use when users want to clone their voice, create a custom voice, or ask "怎么克隆声音", "我想用自己的声音合成", "音色克隆怎么做". Walks users through recording requirements, checks audio quality before they upload, guides them to the platform for cloning, then generates a preview with their new voice.
metadata
{"openclaw":{"requires":{"env":["SENSEAUDIO_API_KEY"],"bins":["curl","jq","xxd"]},"primaryEnv":"SENSEAUDIO_API_KEY","homepage":"https://senseaudio.cn"}}
compatibility
{"required_credentials":[{"name":"SENSEAUDIO_API_KEY","description":"API key from https://senseaudio.cn/platform/api-key","env_var":"SENSEAUDIO_API_KEY"}]}

SenseAudio Clone Wizard

A guided experience for voice cloning. The API itself is simple — the hard part is getting a clean recording. This skill handles that.

Note: SenseAudio does not support cloning via API. The actual cloning happens on the platform at https://senseaudio.cn. This skill guides the user through the full process and handles the steps that can be automated.


Phase 1: Before Recording

When a user says they want to clone their voice, show this guide before they record anything:


录音前请注意:

  • 环境:找一个安静的房间,关掉空调、风扇、电视
  • 距离:嘴巴距离麦克风约 20cm,不要太近(会爆音)也不要太远(声音发虚)
  • 时长:3–30 秒,朗读一段自然的文字效果最好
  • 格式:MP3、WAV 或 AAC,文件不超过 50MB
  • 内容:朗读一段话,语速自然,避免长时间停顿

推荐朗读内容(约 15 秒):

"大家好,我是 [你的名字]。今天天气不错,阳光明媚。我喜欢在这样的日子里出去走走,感受大自然的美好。希望每一天都能这样轻松愉快。"

录好后把文件发给我,我来帮你检测质量。


Phase 2: Audio Quality Check

When the user uploads an audio file, run the quality check:

RESULT=$(curl -s -X POST https://api.senseaudio.cn/v1/audio/analysis \
  -H "Authorization: Bearer $SENSEAUDIO_API_KEY" \
  -F "model=sense-asr-check" \
  -F "file=@<AUDIO_FILE>")

Parse the response and translate it into plain language:

Noise score interpretation:

noise_score What to tell the user
0.0 – 0.2 录音质量很好,可以直接提交克隆
0.2 – 0.4 录音质量不错,有轻微噪声,克隆效果可能略有影响,建议在条件允许时重录
0.4 – 0.6 录音噪声较明显,建议重录以获得更好的克隆效果
0.6 – 1.0 录音噪声严重,强烈建议重录,否则克隆出的音色可能失真

Noise type translations:

noise_types value 告诉用户
background_noise 背景噪声(周围环境声)
hum 嗡嗡声(可能是空调、电器)
static 静电杂音(可能是线材或设备问题)
wind 风噪(话筒离嘴太近或有风)
echo / reverb 回声/混响(房间太空旷,建议换小房间或靠近软装)

Example diagnosis output:

你的录音检测结果:
- 时长:14.3 秒 ✓
- 噪声评分:0.52(中等)
- 检测到:嗡嗡声(可能是空调)

建议:关掉空调后重新录一段。如果条件不允许,可以继续提交,
但克隆出的音色可能带有轻微底噪。要继续还是重录?

If has_noise is false or score < 0.2, skip the warning and proceed directly to Phase 3.

Phase 3: Platform Cloning

The cloning itself must be done on the SenseAudio platform — the API does not support direct cloning requests.

Guide the user:


录音质量通过!接下来去平台完成克隆:

  1. 打开 https://senseaudio.cn,登录你的账号
  2. 进入「音色克隆」页面
  3. 点击「上传音频」,选择你刚才的录音文件
  4. 系统会自动训练,通常几秒内完成
  5. 训练完成后,在「我的音色」中找到新音色,复制它的 voice_id
  6. 把 voice_id 发给我,我来帮你生成第一段试听

注意槽位: 每个账号有有限的克隆槽位。如果槽位已满,需要先删除旧音色才能克隆新的。


Phase 4: First Playback

Once the user provides their voice_id, synthesize a welcome message so they can hear their AI voice for the first time:

Default preview text (warm and personal):

"你好!这是我的 AI 声音。从今天起,我可以用这个声音说任何我想说的话了。"

curl -s -X POST https://api.senseaudio.cn/v1/t2a_v2 \
  -H "Authorization: Bearer $SENSEAUDIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d "{
    \"model\": \"SenseAudio-TTS-1.0\",
    \"text\": \"你好!这是我的 AI 声音。从今天起,我可以用这个声音说任何我想说的话了。\",
    \"stream\": false,
    \"voice_setting\": { \"voice_id\": \"<VOICE_ID>\" },
    \"audio_setting\": { \"format\": \"mp3\" }
  }" -o preview.json

jq -r '.data.audio' preview.json | xxd -r -p > my_voice_preview.mp3

After generating: tell the user the file is ready, and mention they can now use this voice_id with any SenseAudio TTS feature — including the senseaudio-tts-quick skill for quick synthesis or senseaudio-polyglot-tts for precise pronunciation control.