CamoVoice User Guide

Fully Private, Offline Speech-to-Text

Overview

CamoVoice is a desktop speech-to-text application designed with one core principle: your voice stays on your device. Unlike cloud-based transcription services, CamoVoice processes everything locally using bundled AI models — no internet connection required, no data leaves your computer, and absolutely zero telemetry.

Privacy & Architecture

100% Offline: All transcription happens on your machine using faster-whisper, which runs entirely locally
Zero Telemetry: CamoVoice collects nothing. No usage analytics, no crash reports, no audio samples
No Cloud Dependencies: Once installed, the app works without any network connection
Local Settings: Your preferences are stored in a simple settings.json file on your device and auto-loaded at open
English Optimized: Purpose-built for English transcription with maximum accuracy

This architecture makes CamoVoice ideal for transcribing sensitive notes, legal dictation, personal journals, or any audio where privacy matters.

User Interface

CamoVoice features a high-contrast, dark-themed interface designed for simplicity and accessibility.

Main Window Layout

┌─────────────────────────────────────────────────────────────┐
│  [Load Audio File]          🎙 Record Button     Status     │
│  [Settings]        "Click or hold spacebar"     🎤 Device   │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│                   Transcription Output                      │
│                      (editable text)                        │
│                                                             │
├─────────────────────────────────────────────────────────────┤
│  [Copy Text] [▶ Play] [Save As]       [Undo Clear] [Clear] │
├─────────────────────────────────────────────────────────────┤
│  [Input Audio: ████████░░] ← (appears during recording)     │
├─────────────────────────────────────────────────────────────┤
│  Text Size: ═══════●═══════    Mode: [Fast|Thinking]        │
└─────────────────────────────────────────────────────────────┘

Recording Audio

There are two ways to record:

Click the Microphone Button: Click once to start, click again to stop
Hold Spacebar: Hold spacebar to record, release to stop (push-to-talk style)
- A brief 200ms delay prevents accidental recordings from quick taps
- If you're typing in the text area, spacebar works normally

During recording, the microphone button pulses and fills with the accent color. When you stop, transcription begins automatically.

Recording Duration Timer

For longer recordings, a duration timer appears after 1 minute of recording, showing your elapsed time and the mode's limit (e.g., 3:45 / 10:00). The timer changes color as you approach the limit:

Gray: Normal — plenty of time remaining
Orange: Warning — less than 1 minute remaining
Red: Critical — less than 30 seconds remaining

This helps you pace your recordings and avoid hitting the automatic cutoff unexpectedly.

Transcription Animation

When CamoVoice is processing your audio, you'll see an animated "Transcribing..." status with cycling dots, and the text area border will pulse with the accent color. This provides clear visual feedback that your audio is being processed.

Audio Level Indicator

While recording, CamoVoice displays a thin Input Audio meter below the action buttons that shows real-time audio levels:

Orange gradient bar: Shows current audio input level
Yellow peak indicator: Marks the highest audio peaks
Automatic display: Appears when recording starts
Smart hiding: Disappears 1.5 seconds after recording stops

This visual feedback confirms your microphone is working and picking up audio.

Loading Audio Files

Click Load Audio File to transcribe existing audio files. CamoVoice supports:

M4A — M4A audio files
WAV — Any sample rate or bit depth
MP3 — Standard MP3 files

The app automatically resamples to the required 16kHz and converts stereo to mono. Files currently have a 60 MB (Fast mode) and 30 MB (Thinking mode) size limit to prevent freezing.

Transcription Output

The large text area displays your transcriptions. Each new recording or file appends to existing text, making it easy to build up longer documents.

Working with Text

Edit freely: Click into the text area to make corrections or additions
Copy Text: Copies all text to your clipboard (button flashes green to confirm)
▶ Play: Reads the transcription aloud using text-to-speech (see Playback Voice below)
Save As: Opens a save dialog with timestamp options (see below)
Undo Clear: Restores the most recently cleared text (grayed out when nothing to undo)
Clear: Removes all text and timestamps

Undo Clear

If you accidentally clear your text, you have two options:

Click the Undo Clear button (located to the left of Clear)
Press Ctrl+Z (Windows) or ⌘Z (macOS)

This restores both the text and any associated timestamps from the most recent clear action. The Undo Clear button is grayed out when there's nothing to restore.

Save As Options

When you click Save As (or press Ctrl+S / ⌘S), a dialog appears with options:

Include date at top of document: Adds a header with the current date
Include timestamp after each transcription: Adds a [Recorded: DD-Mon-YYYY HH:MM:SS] line after each transcription segment (e.g., [Recorded: 09-Jan-2026 14:32:15])
Include edits to transcription: Saves the text exactly as it appears in the app, including any spelling corrections or edits you've made. When this is enabled, timestamps are disabled since edited text may no longer match the original transcription segments.

Export Formats

CamoVoice supports three export formats:

Format	Extension	Best For
Text File	`.txt`	Universal compatibility, email, plain text editors
Word Document	`.docx`	Professional documents, further editing in Microsoft Word
PDF Document	`.pdf`	Sharing, archiving, printing (read-only format)

Simply select your desired format from the "Save as type" dropdown in the file dialog.

Example output with both timestamp options enabled:

Transcription Record - 09-Jan-2026
========================================

This is my first transcription segment.
[Recorded: 09-Jan-2026 14:32:15]

This is my second transcription segment.
[Recorded: 09-Jan-2026 14:35:42]

Transcription Modes

CamoVoice offers two transcription modes, selectable at the bottom of the window:

Model Mode	Speed	Accuracy	Best For
Fast	★★★★★	★★★☆☆	Quick notes, short recordings, real-time feel
Thinking	★★★☆☆	★★★★★	Important transcriptions, longer recordings

Fast mode: Uses the compact base model (~140MB) with greedy decoding for maximum speed. Great for quick voice memos, short dictation, or when you want near-instant results. Trades some accuracy for responsiveness.
Thinking mode: Uses the full large model (~1.4GB) with beam search and voice activity detection. Better accuracy for complex vocabulary, accents, background noise, or professional transcription work.

Your mode selection is automatically saved. The model loads on first use of each mode, so the first transcription after opening the app or switching modes may take slightly longer.

Recordings have a time limit (currently 10 minutes for Fast, 5 minutes for Thinking) to prevent freezing; the recording will be automatically stopped and transcribed when the limit is reached. Break up recordings to avoid limits.

Settings

Click the Settings button to open the settings panel.

Input Device

Select which microphone to use for recording. Options include:

Default: Uses your system's default input device
All detected input devices are listed with their audio API

If you have multiple microphones (e.g., headset, webcam, USB mic), select the one you want to use here.

Playback Voice

Choose the voice used for the ▶ Play feature, which reads your transcription aloud.

These are system voices installed on your device, not voices bundled with CamoVoice. To add or remove voices:

Windows: Settings → Time & Language → Speech → Manage voices
macOS: System Settings → Accessibility → Spoken Content → System Voice → Manage Voices

The available voices depend on what you've installed on your operating system.

Show Timestamps in Transcriptions

When enabled, each transcription segment displays its recording time directly in the app, formatted like a 24-hour ISO-style legal/medical record with unambiguous month abbreviations:

Your transcribed text here.
[Recorded: 09-Jan-2026 14:32:15]

This is useful for keeping track of when recordings were made. Timestamps are always stored internally — this setting just controls whether they're visible in the app. When you save, you can choose whether to include timestamps regardless of this display setting.

Keep Window Always on Top

When enabled, the CamoVoice window stays visible above other applications. This is useful for:

Dictating while reading documents or browsing
Keeping CamoVoice accessible during multitasking
Hands-free workflows where you can't easily switch windows

Toggle this in Settings → "Keep window always on top".

Custom Words

CamoVoice allows you to add custom vocabulary — specialized terms, names, phrases, jargon, or unusual words that you frequently use. This helps the speech recognition model recognize these words more accurately.

To manage custom words:

Open Settings
Click Custom Words (n) (where n is your current word count)
In the Custom Words window:
- Type a word or phrase and click Add (or press Enter)
- View all your current custom words in the scrollable list
- Click the ✕ next to any word to remove it
- Click Clear All to remove all custom words
- Click Done when finished

Important notes:

Custom words are suggestions that increase recognition likelihood — they are not guaranteed to be transcribed exactly
You can add up to 100 custom words
Words are case-preserved but duplicate-checked case-insensitively
Custom words are saved automatically and persist between sessions

Best uses for custom words:

Proper names (people, companies, products)
Technical jargon or industry-specific terms
Acronyms that should be recognized as spoken
Unusual spellings or foreign words used in English context

Accessibility Features

Scalable Interface

The Text Size slider adjusts the font size throughout the entire application — not just the transcription area. This includes:

Transcription text
All buttons and labels
Status messages
Mode selector

Slide right for larger text, left for smaller. Your preference is automatically saved.

This makes CamoVoice usable for:

Users with visual impairments
High-DPI displays
Presentations or screen sharing
Personal comfort preferences

Keyboard Shortcuts

Action	Windows	macOS
Hold to record	Spacebar (hold)	Spacebar (hold)
Save As	Ctrl+S	⌘S
Undo clear	Ctrl+Z	⌘Z
Close dialogs	Escape	Escape

Note: The Escape key closes Settings and Save As dialogs without saving changes.

How It Works (Technical)

CamoVoice is built on several key technologies:

Speech Recognition

Two model options: Fast mode uses base.en (~140MB), Thinking mode uses distil-large-v3 (~1.4GB)
faster-whisper: CTranslate2-based implementation that's 4-6x faster than the original Whisper with lower memory usage
Models are bundled with the application — no download required after installation
All processing uses your CPU with int8 quantization for efficiency

Audio Capture

SoundDevice: Cross-platform audio recording
Records at 16kHz mono (Whisper's native sample rate)
Audio is processed directly in memory — no temporary files

Text-to-Speech

pyttsx3: Interfaces with your operating system's speech synthesis
Uses voices installed on your system (SAPI5 on Windows, NSSpeechSynthesizer on macOS)

User Interface

CustomTkinter: Modern-looking Python GUI framework
Dark theme with orange accent colors
Responsive layout that scales with window size (minimum 640×480)

Troubleshooting

"No voices available" for Playback

Your system may not have TTS voices installed. Install voices through your operating system's speech settings.

Recording doesn't start

Check that your microphone is connected and selected in Settings
Ensure no other application is exclusively using the microphone
Try selecting a different Input Device

Transcription is inaccurate

Speak clearly and at a moderate pace
Reduce background noise
Try switching to Thinking mode for better accuracy
Ensure you're speaking English (CamoVoice is optimized for English)

App is slow on first transcription

The first transcription may take slightly longer as the model initializes. Subsequent transcriptions will be faster. Switching modes also requires loading a different model.

Tips for Best Results

Speak clearly with natural pacing — the model handles conversational speech well
Minimize background noise — even with noise handling, clearer audio = better results
Use Fast mode for quick notes or short sequential dictations, Thinking mode for important uninterrupted long transcriptions
Edit as you go — freely adjust the text manually to fix mistakes or add/remove text before or after transcriptions

Enterprise Customization & Expansion Options

17 input languages: English, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Chinese, Japanese, Korean, Polish, Turkish, Ukrainian, Arabic, Hindi, and Vietnamese
Option to auto-detect input language
Option to auto-translate to English
Additional file formats for input sounds and output files
Enhanced audit trail features and export formats

Privacy Commitment

✓ No internet connection required after installation
✓ No accounts, no sign-ups, no authentication
✓ No analytics or usage tracking
✓ No audio sent to any server
✓ Settings stored locally in plain JSON
✓ Models bundled locally — no APIs, no third party exposure

CamoVoice v0.1.1