AgentSkillsCN

agentvibes

Autoaccl组件生成器

SKILL.md
--- frontmatter
name: agentvibes
description: Text-to-speech using AgentVibes with PulseAudio/SSH tunnel support for remote servers. Use when you need to generate speech audio that plays on a local machine from a remote server, or when free/offline TTS is preferred over cloud APIs. Provides Piper TTS (50+ voices, multilingual, 100% free) with automatic audio tunneling through SSH.

AgentVibes TTS for Clawdbot

Generate text-to-speech audio using AgentVibes with full support for remote SSH sessions and PulseAudio tunneling.

Quick Start

Basic TTS

bash
# Simple TTS with default voice
npx agentvibes speak "Your text here"

# With a specific voice
npx agentvibes speak "Hello" --voice en_US-amy-medium

Choose Your Voice

bash
# List all 50+ available voices
npx agentvibes voices

# Set your default voice
npx agentvibes voice set en_US-amy-medium

Popular Voice Options

  • Female voices: en_US-amy-medium, en_GB-jenny_dioco-medium, fr_FR-siwis-medium
  • Male voices: en_US-lessac-medium (default), en_GB-alan-medium, en_US-ryan-high
  • Other languages: es_ES-davefx-medium (Spanish), fr_FR-siwis-medium (French), de_DE-thorsten-medium (German)

Features

Core Features

  • Free & Offline: Uses Piper TTS (no API keys required)
  • 50+ Voices: Professional AI voices in 30+ languages
  • Remote SSH Support: Audio automatically tunnels to your local machine via PulseAudio
  • Zero Latency: Local generation, instant playback
  • Integrated: Works seamlessly with Clawdbot messaging

Advanced Features (All Available!)

  • 🎵 Background Music: 50+ pre-made tracks (jazz, lofi, flamenco, cumbia, bossa nova, etc.)
  • 🎚️ Voice Effects: Reverb (light, medium, heavy, cathedral)
  • 🎭 Personalities: 19+ styles (pirate, sarcastic, zen, flirty, robot, therapist, etc.)
  • 🌍 Multi-language: Auto-translation and language learning modes
  • ⚡ Customization: Speed control, sentiment, verbosity levels

Installation

bash
npx agentvibes install

That's it! AgentVibes is ready to use.

For remote SSH audio, see: https://github.com/paulpreibisch/AgentVibes/blob/master/docs/remote-audio-setup.md

Usage Examples

Basic TTS

bash
agentvibes speak "Hello from Clawdbot"

Voice Selection

AgentVibes provides 50+ professional AI voices. Users can select voices in multiple ways:

Method 1: One-Time Voice (Per Message)

bash
# Use specific voice for this message only
bash ~/.claude/hooks/play-tts.sh "Hello world" "en_US-amy-medium"

# Different voices:
bash ~/.claude/hooks/play-tts.sh "Hello" "en_US-lessac-medium"  # Default male (US)
bash ~/.claude/hooks/play-tts.sh "Hello" "en_US-amy-medium"     # Female (US)
bash ~/.claude/hooks/play-tts.sh "Hello" "en_US-ryan-high"      # High quality male (US)
bash ~/.claude/hooks/play-tts.sh "Good day" "en_GB-alan-medium" # British male
bash ~/.claude/hooks/play-tts.sh "Bonjour" "fr_FR-siwis-medium" # French female
bash ~/.claude/hooks/play-tts.sh "Hola" "es_ES-davefx-medium"   # Spanish male

Method 2: Set Default Voice (Persistent)

bash
# List all available voices with preview samples
bash ~/.claude/hooks/voice-manager.sh list

# Set your preferred voice as default
bash ~/.claude/hooks/voice-manager.sh set en_US-amy-medium

# Now all messages use this voice
bash ~/.claude/hooks/play-tts.sh "This uses Amy's voice"

# Check current default voice
bash ~/.claude/hooks/voice-manager.sh show

Method 3: Preview Before Choosing

bash
# Preview a voice before committing
bash ~/.claude/hooks/voice-manager.sh preview en_US-ryan-high

# Preview multiple voices to compare
bash ~/.claude/hooks/voice-manager.sh preview en_GB-alan-medium
bash ~/.claude/hooks/voice-manager.sh preview en_US-lessac-medium

# Set your favorite after testing
bash ~/.claude/hooks/voice-manager.sh set en_GB-alan-medium

Voice Categories

🇺🇸 English (US) - Most Popular:

  • en_US-lessac-medium - Clear male voice (default)
  • en_US-amy-medium - Friendly female voice
  • en_US-ryan-high - High quality male voice
  • en_US-libritts-high - Multiple speakers available

🇬🇧 English (UK):

  • en_GB-alan-medium - British male
  • en_GB-jenny_dioco-medium - British female

🌍 Other Languages:

  • es_ES-davefx-medium - Spanish (Spain) male
  • es_MX-claude-high - Spanish (Mexico) male
  • fr_FR-siwis-medium - French female
  • fr_FR-gilles-low - French male
  • de_DE-thorsten-medium - German male
  • de_DE-eva_k-x_low - German female
  • it_IT-riccardo-x_low - Italian male
  • pt_BR-faber-medium - Portuguese (Brazilian) male
  • ja_JP-ayanami-medium - Japanese female
  • zh_CN-huayan-x_low - Chinese female
  • ko_KR-kss-medium - Korean female

See full list: https://rhasspy.github.io/piper-samples/

Advanced Features

AgentVibes provides powerful audio customization through hook scripts:

🎵 Background Music

bash
# Enable background music
bash ~/.claude/hooks/background-music-manager.sh on

# List available tracks
bash ~/.claude/hooks/background-music-manager.sh list

# Set specific track (flamenco, jazz, lofi, etc.)
bash ~/.claude/hooks/background-music-manager.sh set-default agentvibes_soft_flamenco_loop.mp3

# Adjust volume (0.0-1.0)
bash ~/.claude/hooks/background-music-manager.sh volume 0.3

# Disable background music
bash ~/.claude/hooks/background-music-manager.sh off

🎚️ Voice Effects (Reverb)

bash
# Add reverb effects
bash ~/.claude/hooks/effects-manager.sh set-reverb light      # Small room
bash ~/.claude/hooks/effects-manager.sh set-reverb medium     # Conference room
bash ~/.claude/hooks/effects-manager.sh set-reverb heavy      # Large hall
bash ~/.claude/hooks/effects-manager.sh set-reverb cathedral  # Epic space

# Turn off reverb
bash ~/.claude/hooks/effects-manager.sh set-reverb off

# Check current effects
bash ~/.claude/hooks/effects-manager.sh list

🎭 Personalities

Add personality and emotion to the voice:

bash
# Set personality (sarcastic is hilarious!)
bash ~/.claude/hooks/personality-manager.sh set sarcastic
bash ~/.claude/hooks/personality-manager.sh set flirty
bash ~/.claude/hooks/personality-manager.sh set zen
bash ~/.claude/hooks/personality-manager.sh set robot

# List all personalities
bash ~/.claude/hooks/personality-manager.sh list

# Remove personality
bash ~/.claude/hooks/personality-manager.sh unset

Available personalities: sarcastic (highly recommended!), flirty, pirate, robot, zen, valley-girl, noir-detective, wizard, shakespearean, drill-sergeant, therapist, game-show-host, conspiracy-theorist, stoner, dry-humor, enthusiastic, professional, drunk, and more!

🗣️ Voice & Speed Control

bash
# List all 50+ available voices
bash ~/.claude/hooks/voice-manager.sh list

# Preview a voice before selecting
bash ~/.claude/hooks/voice-manager.sh preview en_US-amy-medium

# Set default voice (persists across messages)
bash ~/.claude/hooks/voice-manager.sh set en_US-amy-medium

# Check current voice
bash ~/.claude/hooks/voice-manager.sh show

# Use specific voice for one message only
bash ~/.claude/hooks/play-tts.sh "Hello" "en_GB-alan-medium"

# Adjust speech speed (0.5-2.0)
bash ~/.claude/hooks/speed-manager.sh set 1.2  # 20% faster
bash ~/.claude/hooks/speed-manager.sh set 0.8  # 20% slower
bash ~/.claude/hooks/speed-manager.sh show     # Check current speed

Popular Voice Choices:

  • Female: en_US-amy-medium, en_GB-jenny_dioco-medium, fr_FR-siwis-medium
  • Male: en_US-lessac-medium, en_GB-alan-medium, en_US-ryan-high
  • High Quality: en_US-ryan-high, es_MX-claude-high

🗣️ Verbosity Control

Control how much detail gets spoken via TTS:

bash
# Set verbosity level
bash ~/.claude/hooks/verbosity-manager.sh set high      # Maximum transparency (recommended for Clawdbot)
bash ~/.claude/hooks/verbosity-manager.sh set medium    # Balanced (major decisions only)
bash ~/.claude/hooks/verbosity-manager.sh set low       # Minimal (acknowledgments only)

# Check current level
bash ~/.claude/hooks/verbosity-manager.sh get

# Get info about levels
bash ~/.claude/hooks/verbosity-manager.sh info

Verbosity Levels:

  • HIGH - All reasoning spoken (full transparency) - Recommended for Clawdbot
  • MEDIUM - Major decisions and findings only
  • LOW - Just acknowledgments and completions

For Clawdbot: Set to HIGH for full verbosity in all responses.

🌍 Language & Translation

bash
# Set target language
bash ~/.claude/hooks/language-manager.sh set-language es  # Spanish
bash ~/.claude/hooks/language-manager.sh set-language fr  # French

# Enable auto-translation
bash ~/.claude/hooks/translate-manager.sh on

# Disable translation
bash ~/.claude/hooks/translate-manager.sh off

Complete Example with Effects

bash
# Setup: Flamenco music + cathedral reverb + sarcastic personality
bash ~/.claude/hooks/background-music-manager.sh set-default agentvibes_soft_flamenco_loop.mp3
bash ~/.claude/hooks/effects-manager.sh set-reverb cathedral
bash ~/.claude/hooks/personality-manager.sh set sarcastic

# Speak with all effects applied
bash ~/.claude/hooks/play-tts.sh "Oh great, another user who needs help. How wonderful."

Programmatic Use

bash
# Capture audio file path
OUTPUT=$(agentvibes speak "Test")
AUDIO_FILE=$(echo "$OUTPUT" | grep "Saved to:" | cut -d: -f2- | xargs)
echo "Audio: $AUDIO_FILE"

Voice Library

Common voices:

  • English (US): en_US-lessac-medium (default male), en_US-amy-medium (female)
  • English (UK): en_GB-alan-medium (male), en_GB-jenny_dioco-medium (female)
  • Spanish: es_ES-davefx-medium (male), es_MX-claude-high (female HQ)
  • French: fr_FR-siwis-medium (female), fr_FR-gilles-low (male)
  • German: de_DE-thorsten-medium (male), de_DE-eva_k-x_low (female)

See full voice library: agentvibes list-voices

Or browse online: https://rhasspy.github.io/piper-samples/

Remote SSH Audio

When running Clawdbot on a remote server, audio can play on your local machine through SSH tunneling.

Quick setup:

  1. Configure PulseAudio on remote server:
bash
mkdir -p ~/.config/pulse
cat > ~/.config/pulse/default.pa << 'EOF'
.include /etc/pulse/default.pa
load-module module-native-protocol-tcp auth-ip-acl=127.0.0.1;192.168.0.0/16
EOF

echo 'export PULSE_SERVER=tcp:localhost:14713' >> ~/.bashrc
source ~/.bashrc
  1. Configure SSH tunnel on local machine (~/.ssh/config):
code
Host your-server
    RemoteForward 14713 localhost:14713
  1. Connect and test:
bash
ssh your-server
agentvibes speak "Testing remote audio"

Full guide: https://github.com/paulpreibisch/AgentVibes/blob/master/docs/remote-audio-setup.md

Troubleshooting

No audio output

Check PulseAudio:

bash
pactl info

Should show: Server String: tcp:localhost:14713

AgentVibes command not found

Make sure AgentVibes is installed:

bash
npx agentvibes install

Or globally:

bash
npm install -g agentvibes

Audio plays on server instead of local

Verify environment variable:

bash
echo $PULSE_SERVER

Should be: tcp:localhost:14713

Quick Reference - Common Commands

Voice Selection

bash
bash ~/.claude/hooks/voice-manager.sh list                    # List all voices
bash ~/.claude/hooks/voice-manager.sh preview en_US-amy-medium # Preview voice
bash ~/.claude/hooks/voice-manager.sh set en_US-amy-medium     # Set default
bash ~/.claude/hooks/play-tts.sh "Hi" "en_US-amy-medium"       # One-time use

Background Music

bash
bash ~/.claude/hooks/background-music-manager.sh on            # Enable
bash ~/.claude/hooks/background-music-manager.sh list          # List tracks
bash ~/.claude/hooks/background-music-manager.sh set-default agentvibes_soft_flamenco_loop.mp3
bash ~/.claude/hooks/background-music-manager.sh volume 0.3    # Set volume
bash ~/.claude/hooks/background-music-manager.sh off           # Disable

Voice Effects

bash
bash ~/.claude/hooks/effects-manager.sh set-reverb medium      # Add reverb
bash ~/.claude/hooks/effects-manager.sh set-reverb off         # Remove
bash ~/.claude/hooks/effects-manager.sh list                   # Show effects

Personality

bash
bash ~/.claude/hooks/personality-manager.sh list               # List all
bash ~/.claude/hooks/personality-manager.sh set sarcastic      # Set personality
bash ~/.claude/hooks/personality-manager.sh unset              # Remove

Speed Control

bash
bash ~/.claude/hooks/speed-manager.sh set 1.2                  # 20% faster
bash ~/.claude/hooks/speed-manager.sh set 0.8                  # 20% slower
bash ~/.claude/hooks/speed-manager.sh show                     # Current speed

Verbosity Control

bash
bash ~/.claude/hooks/verbosity-manager.sh set high             # Full detail (recommended)
bash ~/.claude/hooks/verbosity-manager.sh set medium           # Balanced
bash ~/.claude/hooks/verbosity-manager.sh set low              # Minimal
bash ~/.claude/hooks/verbosity-manager.sh get                  # Check current

Technical Details

  • Audio Format: WAV (16-bit, 22.05kHz mono)
  • Storage: ~/.claude/audio/tts-processed-*.wav
  • Engine: Piper TTS (VITS neural TTS from Hugging Face)
  • Latency: <1 second for typical sentences

Documentation