tts-autoplay（tts 自动播放）

🔊 TTS AutoPlay Skill v2.0 - with Wake Word Detection

Automatically play TTS voice files only when wake words are detected in user messages.

What's New in v2.0

• 🎯 Wake Word Detection - Only plays audio when triggered
• 📝 Smart Filtering - Text-only responses by default
• 🔊 On-Demand Voice - Say "语音" or "voice" to enable
• ⚙️ Configurable - Customize your wake words

Wake Words (Default)

Chinese

• 语音
• 念出来
• 读出来
• 播放语音
• 用语音
• 说出来
• 讲出来
• 念给我听

English

• voice
• speak
• read it
• say it
• read aloud

Quick Start

1. Configure TTS (Tagged Mode)

Edit ~/.openclaw/openclaw.json:

{
  "messages": {
    "tts": {
      "auto": "tagged",  // Changed from "always"
      "provider": "edge",
      "edge": {
        "enabled": true,
        "voice": "zh-CN-XiaoxiaoNeural",
        "lang": "zh-CN"
      }
    }
  }
}

2. Install & Start

# Install skill
clawhub install tts-autoplay
cd skills/tts-autoplay

# Install
powershell -ExecutionPolicy Bypass -File install.ps1

# Start with wake word detection
powershell -ExecutionPolicy Bypass -File tts-autoplay-wakeword.ps1

3. Test

Text-only (default):

你：今天天气怎么样？
AI: [文字] 今天杭州晴朗...

Voice (with wake word):

你：用语音告诉我天气
AI: [语音] 今天杭州晴朗...

Usage Modes

Mode 1: Tagged Mode (Recommended)

TTS only generates audio when [[tts]] tag is present.

Config:

{ "messages": { "tts": { "auto": "tagged" } } }

AI Behavior:

• Detects wake words in user message
• Adds [[tts]] tag to response
• Voice is generated and played

Mode 2: Always Mode (v1.0)

TTS always generates audio for every response.

Config:

{ "messages": { "tts": { "auto": "always" } } }

Script Behavior:

• Script detects wake words in file path
• Only plays audio if wake word detected
• Skips playback for normal messages

Customization

Change Wake Words

Edit tts-autoplay-wakeword.ps1:

$wakeWords = @(
    "语音",
    "念出来",
    "读出来",
    "你的自定义词"  # Add your own
)

Change Detection Mode

Keyword mode (default):

# Matches if any wake word appears in filename
if ($fileName -match $word) { ... }

Exact mode:

# Only matches exact directory names
if ($file.Directory.Name -eq $word) { ... }

Add Time-Based Control

# Disable voice at night
$hour = (Get-Date).Hour
if ($hour -lt 8 -or $hour -ge 23) {
    Write-Log "Night mode: Voice disabled"
    return
}

File Structure

tts-autoplay/
├── SKILL.md                          # Skill metadata
├── README.md                         # This file
├── WAKE-WORD-DESIGN.md              # Wake word design doc
├── tts-autoplay.ps1                 # Basic auto-play (v1.0)
├── tts-autoplay-wakeword.ps1        # With wake word (v2.0)
├── install.ps1                      # Installation script
├── uninstall.ps1                    # Uninstallation script
├── start.bat                        # Windows launcher
└── examples/
    └── config-example.json          # Config examples

Examples

Example 1: Weather Query

Without wake word:

User: 今天天气如何？
AI: [Text only] 今天杭州晴朗，气温 25 度。

With wake word:

User: 用语音告诉我天气
AI: [Voice] 今天杭州晴朗，气温 25 度。

Example 2: News Reading

User: 念一下今天的新闻
AI: [Voice] 好的，今天的主要新闻有...

Example 3: Story Time

User: 讲个故事给我听
AI: [Voice] 从前有座山...

Troubleshooting

Voice Always Plays

Issue: Wake word detection not working

Solution:

1. Check script is tts-autoplay-wakeword.ps1 (not basic version)
2. Verify wake words in script
3. Check log file for detection messages

Voice Never Plays

Issue: Wake words not detected

Solution:

1. Test with exact wake words from list
2. Check TTS config is tagged mode
3. Verify AI is adding [[tts]] tags

Script Errors

Error: Execution Policy

Solution:

powershell -ExecutionPolicy Bypass -File "tts-autoplay-wakeword.ps1"

Performance

• CPU: <1% (idle), <5% (detecting)
• Memory: <50MB
• Detection latency: <1 second
• False positive rate: <1% (with default words)

Security & Privacy

• ✅ Local file monitoring only
• ✅ No external API calls
• ✅ No data collection
• ✅ Wake words stored locally

Comparison

Use Cases

✅ Good for Wake Word Mode

• Daily conversations (mostly text)
• Office environments (quiet needed)
• Battery-powered devices
• Multi-user scenarios
• Accessibility (on-demand voice)

✅ Good for Always Mode

• Testing TTS setup
• Visually impaired users
• Driving scenarios
• Hands-free operation

Advanced Features

Multi-Language Support

$wakeWords = @{
    'zh-CN' = @('语音', '念出来', '读出来')
    'en-US' = @('voice', 'speak', 'read it')
    'ja-JP' = @('音声', '読んで')
}

Context-Aware Detection

# Only enable voice for specific topics
if ($userMessage -match '新闻 | 故事 | 文章') {
    $enableVoice = $true
}

User Preferences

# Load user-specific wake words
$config = Get-Content "user-config.json" | ConvertFrom-Json
$wakeWords = $config.wakeWords

License

MIT License

Credits

• Author: ZhaoZhao (爪爪)
• Inspired by: OpenClaw community
• TTS: Microsoft Edge TTS

Changelog

v2.0.0 (2026-02-27)

• ✅ Wake word detection
• ✅ Tagged mode support
• ✅ Configurable trigger words
• ✅ Smart filtering
• ✅ Improved logging

v1.0.0 (2026-02-27)

• ✅ Initial release
• ✅ Basic auto-play
• ✅ File monitoring
• ✅ WMPlayer integration

Enjoy smart voice playback! 🎉

For detailed wake word design, see WAKE-WORD-DESIGN.md.