Melo User Guide

When you launch Melo, an icon appears in the menu bar. The app runs in the background, and you only press the hotkey to record when you need voice input.

First-launch permissions

• Microphone permission: When the system asks for access, choose "Allow"
• Accessibility permission: Required to paste text into other apps

Voice input flow

1. Press the hotkey (default: Option + Space)
   • The menu bar icon switches to a recording state
   • An overlay bubble appears
2. Speak
   • Speak in Japanese, English, or a mix of both
   • Text appears in real time on the overlay
3. Press the hotkey again (in Toggle mode)
   • Recording ends and transcription begins
4. Review and edit the text
   • You can correct the text on the overlay
   • Use the "Paste" button to insert it directly into the active app
   • Or the text is automatically copied to the clipboard (paste with Cmd + V)

Toggle mode (default)

Action	Behavior
Press the hotkey	Start recording
Press the hotkey again	End recording → Transcribe

Use it like a push button. Press, speak, then press again to finish.

PTT (Push-to-Talk) mode

Action	Behavior
Hold the hotkey	Recording
Release the hotkey	End recording → Transcribe

Like a transceiver, it records only while you hold the key.

How to set it: Settings → "Recording mode" → select "Push to Talk"

While recording, an overlay bubble is shown on screen.

During recording

• An audio waveform is displayed
• Transcribed text is shown in real time
• The text is scrollable

After recording ends

• The text becomes editable
• You can:
  • Paste button: Insert the text at the cursor position in the active app
  • Manual editing: Correct misrecognized parts

You can register words you use often, or words you want to replace with a specific spelling.

How the dictionary works

• Reading: The text output by speech recognition (before replacement)
• Output: The spelling you actually want (after replacement)

Examples

Reading	Output	Description
アンチグラビティ	antigravity	Convert katakana to an English word
エーアイ	AI	Convert a reading to an abbreviation
ぴーえいちピー	pH	Spelling that matches the pronunciation

Matching rules

• Partial match: Replaces the "reading" wherever it appears in the text
• Longest match: When multiple entries match, the longest one wins
• Case-sensitive

Managing the dictionary

1. Click the menu bar icon
2. Open Settings
3. Select the "Dictionary" tab
4. Add, edit, delete, and search entries

From the menu bar icon → Settings, you can change the following.

Setting	Description	Default
Launch at login	Start Melo automatically when your Mac starts	OFF
Recording hotkey	Key used to start/end recording	Option + Space
Recording mode	Toggle / Push to Talk	Toggle
Cancel hotkey	Cancel recording	Option + Esc
Max recording time	Maximum recording seconds per session	3 min (180 sec)
Language mode	Auto detect / Japanese / English	Japanese
Success notification	Notify on successful transcription	ON
Error notification	Notify on errors	ON

Shortcut	Action	Notes
Option + Space	Start/Stop recording	In Toggle mode
Option + Space (hold)	Recording	In PTT mode
Option + Esc	Cancel recording	Only available while recording
Cmd + V	Paste text	From the clipboard

Custom shortcuts

You can freely change the hotkeys from the Settings screen.

Transcription doesn't start

• Microphone permission: System Settings → Privacy & Security → Microphone → enable Melo
• Model status: On first launch, a model download may be required
• Recording length: If the recording is too short (under 1 second), it may not be recognized

Recognition accuracy is poor

• Change the language mode: Try fixing the language to "Japanese" or "English" in Settings
• Use the user dictionary: Register words that are often misrecognized
• Recording environment: Speaking closer to the mic in a quiet environment improves accuracy

The hotkey doesn't work

• It may conflict with another app's hotkey
• Change it to a different key from the Settings screen

The app won't launch

• macOS 26.2 or later is required
• On first launch, allow "Open this app" in System Settings

Term	Description
STT	Speech-to-Text. Technology that converts speech into text
Whisper	A speech recognition model developed by OpenAI. Melo uses it locally
PTT	Push-to-Talk. A mode that records only while the key is held
Toggle	A mode that switches on/off each time the key is pressed
User dictionary	Word replacement rules defined by the user
Overlay	A UI displayed on top of the screen