Wake Word & Always Listening PRO

Wake word detection enables hands-free voice activation. Say a wake word to start recording without clicking or pressing any keys. Perfect for workflows where your hands are busy or you need truly frictionless voice input.

What is Wake Word Detection?

Wake word detection is an always-on listening mode that monitors your microphone for a specific trigger phrase. When the wake word is detected, StarWhisper automatically starts recording your voice without requiring any manual interaction.

Key Benefits:

• Hands-free activation - No need to click or press hotkeys
• Low CPU usage - Lightweight detection runs efficiently in the background
• Works everywhere - Activate recording from any application
• Natural workflow - Just speak when you need transcription

How It Works

When Always Listening mode is enabled, StarWhisper runs a lightweight wake word detection service using OpenWakeWord, an efficient machine learning model optimized for low-resource environments.

1. Continuous monitoring: The wake word detector listens to your microphone audio
2. Wake word detected: When your chosen wake word is spoken, StarWhisper activates
3. Recording starts: The main recording begins automatically
4. End word support: Say "end transcript" to stop recording hands-free
5. Auto-transcription: Audio is transcribed and copied to clipboard as usual

Available Wake Words

StarWhisper includes several built-in wake words powered by OpenWakeWord. Choose the one that works best for your environment and preferences:

Wake Word	Description	Best For
Alexa	Default wake word	General use, familiar trigger
Computer	Classic voice assistant trigger	Professional environments
Hey Rhasspy	Alternative open-source trigger	Avoiding conflicts with other devices

Custom wake words are supported through OpenWakeWord's model training system, though this requires advanced setup outside of StarWhisper's built-in interface.

Enabling Wake Word Detection

Step 1: Enable Always Listening

1. Right-click the violet recording circle
2. Select Settings from the context menu
3. Navigate to the Wake Word section
4. Enable the Always Listening toggle

Step 2: Choose Your Wake Word

1. Select your preferred wake word from the dropdown menu
2. StarWhisper will automatically load the corresponding detection model
3. Wait for the "Listening for [wake word]..." status message

Step 3: Adjust Sensitivity (Optional)

The Wake Word Sensitivity slider controls how strictly the detector matches your voice:

• Lower sensitivity (0.3-0.5): Reduces false positives, requires clearer pronunciation
• Default (0.6): Balanced for most environments
• Higher sensitivity (0.7-0.9): More responsive, may trigger on similar-sounding words

Using Wake Word Detection

Basic Usage

1. Ensure Always Listening is enabled (status shows "Listening for [wake word]...")
2. Say your chosen wake word (e.g., "Alexa")
3. Wait for the acknowledgment sound
4. Begin speaking your dictation
5. Say "end transcript" to stop, or click the circle

End Word Feature

When recording starts via wake word, you can stop the recording hands-free by saying:

• "end transcript"

This phrase is automatically detected during wake word-initiated recordings, allowing fully hands-free operation from start to finish.

Tips for Best Results

Environment

• Quiet spaces: Wake word detection works best with minimal background noise
• Consistent pronunciation: Say the wake word clearly and consistently each time
• Microphone placement: Position your microphone 6-12 inches from your mouth
• Avoid TV/radio: Media playback can cause false positives

Avoiding False Positives

• Lower sensitivity if wake word triggers accidentally
• Choose a wake word that doesn't appear in normal conversation
• Avoid wake words similar to names of people you talk to
• Temporarily disable Always Listening during calls or meetings

System Requirements

• Python 3.8+ installed on your system (required for OpenWakeWord)
• RAM: Additional 100-150MB for wake word detection service
• CPU: Minimal impact (~1-3% on modern processors)
• Microphone: Always-on microphone access (grant permissions in OS settings)

Troubleshooting

Issue	Solution
Wake word not detected	Increase sensitivity slider, speak louder, check microphone is working
Too many false positives	Decrease sensitivity, choose different wake word, reduce background noise
"Python not found" error	Install Python 3.8+ from python.org and ensure it's in PATH
Wake word service won't start	Check microphone permissions, restart StarWhisper, reinstall Python dependencies
High CPU usage	Restart wake word service (disable/enable Always Listening), check for system conflicts

Privacy & Security

Wake word detection runs entirely locally on your computer. No audio is sent to external servers during wake word monitoring.

• Local processing: All wake word detection happens on-device
• No cloud dependency: Works completely offline
• Minimal data capture: Only detects wake word presence, doesn't record continuously
• User control: Disable anytime via Always Listening toggle

Comparison: Wake Word vs Manual Recording

Feature	Wake Word	Manual Recording
Activation	Voice only	Click or hotkey
Hands-free	Yes	No (requires interaction)
Resource usage	+100MB RAM, +1-3% CPU	Minimal overhead
False activations	Possible (rare)	None
Best for	Hands-busy workflows, cooking, surgery, accessibility	General dictation, controlled recording

Next Steps

• Learn recording fundamentals
• Enable real-time transcription
• Configure custom hotkeys
• Upgrade to PRO to unlock wake word detection