Live Speech-to-Text
How to use Yapper's real-time voice dictation
Live speech-to-text is Yapper’s core feature—press a shortcut, speak, and watch your words appear in any application. This guide covers everything you need to know about live dictation.
How It Works
- You press the recording shortcut (default: ⌥ Space)
- Yapper captures your audio using your Mac’s microphone
- Speech is processed locally using an on-device AI model
- Text is typed into the active application’s text field
The entire transcription happens on your Mac—your voice never leaves your computer.
Transcription Modes
Yapper offers two transcription modes depending on your setup:
Streaming Mode
When using a Whisper model, text appears word-by-word in real time as you speak. You’ll see your words flowing into the active text field while you’re still talking — no need to wait for processing after you stop recording.
Batch Mode
When using Parakeet models, your speech is processed after you stop recording. The overlay shows the processing stage, then the full text is typed at once. Parakeet’s speed (110-190x real-time) makes batch processing feel nearly instant.
Recording Modes
Yapper automatically detects how you use the shortcut and adapts to your style.
Toggle Mode
How to activate: Press and quickly release the shortcut (under 0.3 seconds)
Behavior:
- Recording starts when you press the shortcut
- Recording continues until you press the shortcut again
- No need to hold any keys while speaking
Best for:
- Longer dictations (paragraphs, emails)
- Hands-free use
- When you need both hands for other tasks
Hold-to-Record Mode
How to activate: Press and hold the shortcut for more than 0.3 seconds
Behavior:
- Recording starts after the 0.3-second threshold
- Recording continues while you hold the keys
- Recording stops when you release any key
Best for:
- Quick phrases or sentences
- When you want precise control over recording duration
- Push-to-talk style interaction
Choosing a Mode
You don’t need to configure anything—just use whichever feels natural:
- Quick press → Toggle mode
- Press and hold → Hold-to-record mode
Many users prefer toggle mode for longer dictation and hold-to-record for quick corrections or short inputs.
The Recording Overlay
When you start recording, a small floating indicator appears near your cursor. This overlay shows you what Yapper is doing.
Overlay States
| Color | Icon | Meaning |
|---|---|---|
| Red | Animated waveform | Listening to your voice |
| Blue | Spinner | Processing speech to text |
| Purple | Sparkle | AI is working (Transform or Q&A) |
| Green | Checkmark | Success—text has been typed |
| Red | Warning | Error occurred (message shown) |
The waveform animation while recording confirms Yapper is capturing audio. If you see the waveform moving but get poor results, check your microphone selection in Settings → Preferences.
If you have a secondary language configured and it’s currently active, a flag emoji also appears on the overlay to indicate which language is being used for transcription.
Overlay Position
The overlay appears near your cursor position with a frosted-glass design that adapts to your system’s light or dark mode. It automatically adjusts to stay on screen.
If you prefer a fixed location, you can enable Fixed Position in Settings → Preferences to pin the overlay to the top-center of your screen.
Canceling a Recording
To discard a recording without typing any text:
- Press Escape (default cancel shortcut)
Escape works at any stage—during recording, processing, or AI operations. The overlay will disappear and nothing will be typed. Use this if you:
- Start recording by mistake
- Change your mind about what to say
- Want to discard during processing
- Want to start over
Tips for Best Results
Speaking Tips
- Speak naturally — Don’t over-enunciate or speak robotically
- Maintain consistent volume — Avoid whispering then shouting
- Include punctuation cues — Say “period,” “comma,” or “question mark” if the model doesn’t catch them (accuracy varies by language)
- Pause briefly — A short pause at the end helps the model finalize the transcription
Environment Tips
- Reduce background noise — Close windows, turn off fans
- Position your mic well — Closer is generally better
- Consider a headset — External microphones often outperform built-in ones
Settings to Optimize
- Choose the right model — Parakeet TDT v3 (default) or Large v3 Turbo provide the best balance of speed and accuracy
- Set your language — Match the language setting to what you’re speaking
- Use dual-language switching — If you speak two languages regularly, set a secondary language and toggle with ⇧ ⌥ L instead of changing settings each time
- Add custom vocabulary — Technical terms, names, and acronyms can be added to improve recognition
See Transcription Settings for details.
Where Text Gets Typed
Yapper types text into whatever application and text field is currently active (when Auto-Type is enabled). This means:
- The cursor must be in a text field
- The application must accept keyboard input
- No copy-paste is involved—text is typed character by character
If Auto-Type is disabled, transcriptions are still processed and saved to history, but text is not typed into any app. You can toggle auto-type via the menu bar, Settings → Preferences, or the shortcut ⇧ ⌥ T. See Preferences for details.
Supported Applications
Yapper works with virtually any macOS application:
- Native apps (Mail, Notes, Messages, Pages)
- Web browsers (text fields on websites)
- Electron apps (Slack, Discord, VS Code)
- Terminal applications
- Microsoft Office, Google Docs (in browser)
- Any app that accepts keyboard input
Applications That May Need Extra Setup
Some apps with non-standard text input may require:
- Accessibility permission verification
- Running Yapper before opening the app
- Using the app’s native input method
If text doesn’t appear in a specific app, see Text Injection Issues.
Keyboard Shortcuts
The default shortcuts are:
| Action | Default Shortcut |
|---|---|
| Start/Stop Recording | ⌥ Space |
| Cancel Recording | Escape |
| Language Toggle | ⇧ ⌥ L |
| Auto-Type Toggle | ⇧ ⌥ T |
You can customize these in Settings → Preferences → Keyboard Shortcuts. See Preferences for instructions.
AI-Powered Modes
Yapper also offers two AI-powered modes that build on live recording:
- AI Transform — Select text before recording, then speak an instruction to have AI rewrite it
- AI Q&A — Say “Hey Yapper” followed by a question to get an instant AI answer
These modes activate automatically based on context. See AI Features for details.
Related Topics
- AI Features — AI Transform and Q&A overview
- Transcription Settings — Choose models and languages
- Status Indicators — Full reference for overlay states
- Recording Issues — Troubleshoot recording problems