User Guide

1. Installation

System Requirements

• Windows 11 (64-bit)
• .NET 8.0 Desktop Runtime
• 8GB+ RAM recommended
• GPU (CUDA-compatible) for accelerated processing

Installation Steps

1. Download the latest version from the Download page
2. Extract the downloaded ZIP file
3. Run RocketWhisper.exe
4. On first launch, download the Whisper model you want to use

2. Basic Usage

Speech Recognition via Microphone

1. Launch RocketWhisper
2. Click the record button (microphone icon)
3. Speak into the microphone
4. Click the stop button
5. The recognized text appears in the text area

Speech Recognition from File

1. Click the file select button
2. Choose an audio file (WAV/MP3/FLAC/OGG/M4A/WMA)
3. Recognition processing starts automatically
4. The recognized text appears in the text area

Using Recognition Results

• Copy: Click the copy button to copy to clipboard
• Save: Click the save button to save as a text file
• Auto Copy: When enabled in settings, automatically copies upon recognition completion
• Auto Paste: When enabled in settings, automatically pastes upon recognition completion

3. Settings

Click the gear icon to open the settings window. Detailed configuration is available across 8 tabs.

Model & Language Tab

• Model Selection: Choose the Whisper model to use
• Language: Recognition language (Japanese/English/Chinese/Korean/Auto-detect)
• Punctuation Prompt: Instructs Whisper to include punctuation in the output (recommended)

Input Device Tab

• Microphone Device: Select the microphone to use
• Auto Copy: Automatically copy to clipboard upon recognition completion
• Auto Paste: Automatically paste to the active window upon recognition completion

Hotkey Tab

• Record Start/Stop Hotkey: A shortcut to start recording from any application
• Right Alt (Hold): Hold Right Alt to record, release to stop (double-tap for continuous recording)
• Cancel Hotkey: Key to cancel recording (default: Escape)
• AI Command Hotkey: Key to execute AI processing with text selection + voice instruction (default: Ctrl + Space)

Custom Terms Tab

• Custom Term Registration: Register terms to improve recognition accuracy
• Token Usage: Whisper accepts up to 224 tokens' worth of terms (approximately 50 English words)

Text Processing Tab

• Auto Punctuation: Add punctuation via post-processing after recognition
• Auto Line Breaks: Automatically insert line breaks in long text
• Voice Commands: Enable/disable voice command functionality

Correction Rules Tab

• Auto Correction: Enable/disable automatic correction of misrecognition patterns
• Preset Rules: One-click addition of AI/tech terms, numbers/dates, common errors, and filler removal
• Custom Rules: Add and edit custom correction rules (regex supported)

App-Specific Processing Tab

• Processing Modes: Apply different text processing settings per application
• App Mapping: Assign processing modes to specific applications

AI Processing Tab

• AI Provider: Choose from OpenAI, Anthropic, Groq, Gemini, or Local LLM
• Model Selection: Select the AI model to use
• API Key: Enter the API key for your chosen provider
• Connection Test: Verify that settings are correct

License Tab

• License Type: Free, Personal License, Business License
• License Key: Enter and authenticate your license key

4. Global Hotkeys

Regardless of which application you are using, you can instantly start recording with your configured key combination.

Setup

1. Open the settings window
2. Select the "Hotkey" tab
3. Click the input field
4. Press your desired key combination (e.g., Ctrl + Shift + R)
5. Check the "Enabled" checkbox

How to Use

1. Open any application such as a text editor or web browser
2. Press the configured hotkey to start recording
3. When finished speaking, press the hotkey again to stop recording
4. If Auto Paste is enabled, the recognized text is automatically typed into your application

Right Alt Hold Mode (Push-to-Talk)

An intuitive mode where recording occurs only while the Right Alt key is held down, and stops automatically when released.

Basic Operations

Setup

1. Open the settings window
2. Select the "Hotkey" tab
3. Click the "Right Alt (Hold)" button for the Record Start/Stop hotkey
4. Check "Enabled" and save

Cancel Recording

Press the Escape key during recording to cancel without performing recognition. This works in both Hold mode and continuous recording mode.

• Default Key: Escape
• Customize: Settings → Hotkey tab → Cancel Hotkey
• Enable/Disable: Toggle with the checkbox

5. Batch Processing

Transcribe multiple audio and video files at once. Video files are automatically processed by extracting audio via FFmpeg.

Batch Processing Steps

1. Select "Batch Processing" from the menu
2. Add the files you want to process (multiple selection supported)
3. Choose the output format (Text/SRT/VTT)
4. Click "Start Processing"
5. Wait for processing to complete

Supported Formats

• Audio Input: WAV, MP3, FLAC, OGG, M4A, WMA
• Video Input: MP4, MKV, AVI, MOV, WebM, WMV, FLV
• Output: TXT (plain text), SRT (subtitles), VTT (web subtitles)

* FFmpeg is required for video file transcription. It is included with the installer version.

6. Recognition History

Past recognition results are automatically saved and can be referenced or reused at any time.

Viewing History

1. Click the history button (clock icon)
2. A list of past recognition results is displayed
3. Click an item to view its details

History Operations

• Search: Search past recognition results by keyword
• Copy: Copy the selected result to the clipboard
• Delete: Delete unwanted history entries
• Export: Export history as a file

7. Text Processing (Punctuation & Line Breaks)

Features for automatically inserting punctuation and line breaks into recognition results. There are two approaches.

Punctuation Prompt (Recommended)

Instructs Whisper AI to output recognition results that include punctuation.

• How it works: Passes a sample sentence to Whisper to encourage punctuated output
• Advantage: The AI understands context and places punctuation naturally
• Supported languages: Japanese, English, Chinese, Korean

Auto Punctuation (Post-Processing)

Adds punctuation using rule-based logic after recognition.

• How it works: Detects sentence-ending patterns and inserts periods
• Advantage: Useful as a fallback when Whisper doesn't output punctuation
• Note: Pattern-based mechanical processing that doesn't consider context

Auto Line Breaks

Inserts line breaks at appropriate positions in long text.

• Breaks before conjunctions (e.g., "however", "also", "furthermore")
• Splits sentences longer than 100 characters at sentence boundaries

8. Voice Commands

Perform actions like inserting line breaks or deleting text using voice commands.

Available Commands

Usage Tips

• Speak commands clearly at natural pauses in your speech
• Pausing briefly before and after a command improves recognition accuracy
• Voice commands can be enabled/disabled in the settings

9. Custom Terminology

Register industry-specific terms and proper nouns to improve recognition accuracy.

How to Register

1. Open the settings window
2. Select the "Custom Terms" tab
3. Enter a term and click "Add"

Registration Capacity

There is a limit to the number of custom terms Whisper can accept (up to approximately 224 tokens).

• English: Approximately 50 words
• Japanese: Approximately 100 words

Registration capacity can be monitored in real-time in the settings. When the limit is exceeded, older terms are ignored, so prioritize registering frequently used terms.

10. Misrecognition Correction Rules

Set up rules to automatically correct specific misrecognition patterns. Similar to an IME word dictionary, it replaces recognized text strings with different strings.

How to Add Rules

1. Open the settings window
2. Select the "Correction Rules" tab
3. Check "Auto-correct misrecognition patterns"
4. Click the "Edit Correction Rules..." button
5. Enter the "Search Pattern" and "Replacement String"
6. Click "Add Rule"

Basic Usage (Simple Replacement)

Perform simple string replacements without using regular expressions.

Using Regular Expressions (Advanced Replacement)

Check "Use Regular Expressions" to enable advanced pattern matching.

Regular Expression Symbol Guide:

• \s* : Matches zero or more whitespace characters (spaces, tabs, etc.). For example, "micro\s*soft" matches both "microsoft" and "micro soft".
• $ : Matches the end of a line. "you know$" only matches "you know" at the end of a line, not in the middle of a sentence.

11. App-Specific Processing Modes

Automatically apply different text processing settings depending on which application you are using.

Processing Mode List

Processing modes are presets that bundle settings for punctuation, line breaks, voice commands, and more.

Assigning Modes to Apps

1. Open the settings window
2. Select the "App-Specific Processing" tab
3. Click "Add App"
4. Select the target app (from the list of running apps or enter manually)
5. Choose the processing mode to apply

Usage Examples

• Email client: Auto-insert punctuation and line breaks (Smart mode)
• Code editor: No formatting (Simple mode)
• Notepad: Bullet-point-friendly settings (Custom mode)

12. AI Processing (LLM Integration)

Integrate with external AI (LLM) to automatically convert recognition results to business language, casual tone, or perform summarization and translation. With a local LLM (LM Studio / Ollama), AI processing works even in offline environments.

Supported AI Providers

Setup

1. Open the settings window
2. Select the "AI Processing" tab
3. Choose the provider to use
4. Enter the API key (for Local LLM, only the base URL is needed)
5. Click "Connection Test" to verify

AI Processing Modes

Once AI processing is configured, the following modes become available in app-specific processing.

Local LLM Setup Examples

Use LM Studio or Ollama to utilize AI processing completely offline.

LM Studio

LM Studio Setup:

1. Launch LM Studio
2. Select the model to use
3. In the status bar, choose "Power User" or "Developer" from User / Power User / Developer
4. Navigate to the Developer tab (click the green "Developer" icon in the left menu)
5. If the "Status: Running" toggle at the top of the screen is ON, the server is running
6. When "Running", the API URL (e.g., http://localhost:1234) is displayed
7. To stop the server, click the toggle to switch to "Stopped"

RocketWhisper Setup:

Ollama

Ollama Setup:

1. Install and launch Ollama (the API server starts automatically in the background)
2. Download the desired model: ollama pull gemma3
3. Verify downloaded models: ollama list

RocketWhisper Setup:

13. AI Command Mode

Select text, press the dedicated hotkey, and give a voice instruction — the AI instantly processes the text. Supports translation, summarization, tone adjustment, and any other instruction you can think of.

Basic Usage

1. Select text in any application (Notepad, browser, Word, etc.)
2. Press the AI Command hotkey (default: Ctrl + Space)
3. The status changes to "Waiting for voice instruction..." — give your instruction by voice (e.g., "Translate to Japanese", "Summarize this", "Make it formal")
4. Press the AI Command hotkey again to stop recording
5. The AI processes the text, and the result appears in RocketWhisper's text area

Usage Examples

Hotkey Configuration

1. Open the settings window
2. Select the "Hotkey" tab
3. Click the "AI Command Hotkey" input field
4. Press your desired key combination (default: Ctrl + Space)
5. Check "Enabled"

14. Custom Instructions

Register your most-used AI tasks (translation, business rephrasing, summarization, etc.) to dedicated hotkeys. Just speak and the AI processing is applied automatically. Unlike AI Command Mode, you don't need to say "translate this" each time, making it ideal for repetitive tasks.

Setup

1. Configure an AI provider in the "AI Processing" tab of the settings (local LLM or cloud API)
2. Open the "Hotkeys" tab in the settings and find the "Custom Instructions" section
3. Choose from presets (Translate to English, Business Style, Summary, Grammar Fix) or click "Add" to create a new one
4. Assign a dedicated hotkey
5. Make sure "Enabled" is checked

Basic Usage

1. Press the hotkey assigned to your custom instruction
2. Speak into the microphone (any content in any language)
3. Press the same hotkey again to stop recording
4. Whisper recognizes the audio → the pre-configured AI processing is applied automatically → result is output

Built-in Presets

Adding Custom Instructions

1. Go to Settings → "Hotkeys" tab → "Custom Instructions" section → click "Add"
2. Enter a name (e.g., "Translate to French")
3. Enter a prompt (e.g., "Please translate the following text into French")
4. Assign a dedicated hotkey
5. Click "Save"

You can register up to 20 custom instructions.

15. Recording Indicator

When performing hotkey recording while RocketWhisper is minimized to the system tray, a floating window appears to visually confirm the recording state.

Indicator Features

• Real-time Waveform Display: Visualizes audio levels with 16 animated bars
• Processing Mode Display: Shows the currently applied mode (Smart, Business, etc.) with icon and name
• Non-Focus-Stealing: Does not steal focus from other applications
• Draggable: Can be moved to any position; position is automatically saved

Default Position

Initially appears at the bottom center of the screen (above the taskbar). Once dragged to a new position, it will appear there on subsequent recordings.

16. Voice Launcher

Automatically launches an application when a specific keyword is recognized. Simply say "Notepad" to open Notepad — control apps with just your voice.

Setup

1. Open the settings window and select the "Voice Launcher & Search" tab
2. Check "Enable Voice Launcher"
3. Click "Edit Launcher Settings..."
4. Click "Add" to register a keyword and executable file
5. Click "Save" in the settings window

Registration Fields

Matching Behavior

• Exact Match: Launches only when the recognition result exactly matches the keyword
• Punctuation Ignored: Trailing punctuation marks (commas, periods, !, ?) are automatically stripped
• Case Insensitive: English words are matched regardless of case

17. Voice Search

Simply say "Search for..." during normal voice input to automatically open a Google search in your default browser.

Supported Phrases

Setup

1. Open the settings window
2. Select the "Voice Launcher & Search" tab
3. Check "Enable Voice Search"

18. Tray Icon

Check the application's status via the system tray icon.

Icon States

Tray Icon Actions

• Left Click: Show the main window
• Right Click: Show the context menu

19. Troubleshooting