Skip to content

USER_GUIDE.md

Latest commit

 

History

History
476 lines (313 loc) · 9.2 KB

USER_GUIDE.md

File metadata and controls

476 lines (313 loc) · 9.2 KB

DeskTalk User Guide

🎤 Voice-to-Text Transcription Made Easy

DeskTalk listens when you hold a button, transcribes what you said, and types it where your cursor is.

🚀 Quick Start (5 Minutes)

Step 1: Install DeskTalk

  1. Run DeskTalk_0.2.0_x64-setup.exe
  2. Follow installer prompts
  3. Launch from Start Menu

Step 2: Install FFmpeg

Option A - Winget (Easiest):

winget install Gyan.FFmpeg

Option B - Scoop:

scoop install ffmpeg

Option C - Manual:

  1. Download from https://ffmpeg.org/download.html
  2. Extract to C:\ffmpeg
  3. Add C:\ffmpeg\bin to PATH

Step 3: Get a License

  1. Open DeskTalk → License tab
  2. Click "Buy License" ($100 one-time)
  3. Complete payment via Stripe
  4. Check your email for license key
  5. Paste key in DeskTalk → Click "Activate"

Step 4: Get OpenAI API Key (or use Local Model)

For OpenAI (Recommended for Accuracy):

  1. Go to https://platform.openai.com/api-keys
  2. Create new secret key
  3. Copy the key (starts with sk-...)
  4. Paste in DeskTalk → Transcription tab
  5. Click "Validate"

For Local Model (Offline, Free):

  1. Select "Local Model" in Transcription tab
  2. Choose a model (e.g., base.en)
  3. First use will download the model

Step 5: Set Up Push-to-Talk

  1. Open DeskTalk → General tab
  2. Click dropdown next to "Push-to-Talk Key"
  3. Select a key (e.g., F15, F13, CapsLock)
  4. Click "Save Settings"

Step 6: Start Talking!

  1. Status should show "Running" (green)
  2. Click in any text field
  3. Hold your PTT key → Speak → Release
  4. Your words appear! 🎉

📖 Detailed Setup

General Settings

Push-to-Talk Key

  • Choose an unused key (F13-F24 recommended)
  • Avoid keys you use often (don't use Space, Enter, etc.)
  • Best options:
    • F15 - Rarely used function key
    • F13 - Good if you have extended keyboard
    • CapsLock - If you never use it

Audio Input Device

  • "Default Device" - Uses Windows default mic
  • Or select specific microphone
  • Test by recording and checking quality

Text Output Options

  • Capitalize first letter - Start transcriptions with capital
  • Add space after - Auto-space between transcriptions
  • Type characters instead of pasting ⓘ
    • Slower but works in apps that block Ctrl+V
    • Leave off for best speed

System Options

  • Start with Windows - Launch DeskTalk on boot
  • Start minimized to tray - Hide window on startup
  • Dark mode - Easy on the eyes

Transcription Settings

OpenAI API (Cloud)

Pros:

  • Most accurate transcription
  • Handles accents, background noise
  • Fast processing

Cons:

  • Costs money (~$0.006 per minute)
  • Requires internet
  • Sends audio to OpenAI

Setup:

  1. Get key from https://platform.openai.com/api-keys
  2. Paste in field
  3. Click "Validate" to test
  4. Click "View Usage & Billing" to monitor costs

Cost Example:

  • 1000 minutes = ~$6
  • Average user: $2-5/month

Local Model (Offline)

Pros:

  • Free forever
  • Works offline
  • Private (audio stays on your PC)

Cons:

  • Less accurate than OpenAI
  • Slower on older PCs
  • Larger download (~1GB for best model)

Models:

  • tiny.en - Fastest, least accurate (75 MB)
  • base.en - Good balance (142 MB)
  • small.en - Better accuracy (466 MB)
  • medium.en - Best accuracy (1.5 GB)

License Tab

License Status

  • Active (Green) - Everything works
  • Expired (Yellow) - Renew license
  • Inactive (Red) - Need to activate

Activate License

  1. Paste your license key
  2. Click "Activate"
  3. Key tied to this PC (3 devices max per license)

Deactivate License

  • Click "Deactivate" to free up device slot
  • Useful for switching PCs or testing

Statistics Tab

Track your usage:

  • Total Words - Words transcribed all-time
  • Average WPM - Your speaking speed
  • Transcriptions - Number of recordings
  • Recording Time - Total time speaking

💡 Tips & Tricks

For Best Transcription:

  1. Speak clearly (but naturally)
  2. Use a good microphone (not laptop built-in)
  3. Quiet environment (reduce background noise)
  4. Hold button first, then speak (avoid cutting off start)
  5. Pause briefly before releasing (avoid cutting off end)

Common Use Cases:

Writing Emails:

  • Use PTT for long paragraphs
  • Faster than typing
  • Natural conversational tone

Coding Comments:

  • Quickly document code
  • Explain complex functions
  • Add TODOs

Documentation:

  • Draft documents quickly
  • Brainstorm ideas
  • Meeting notes

Accessibility:

  • RSI/carpal tunnel relief
  • Hands-free input
  • Faster than typing

Keyboard Shortcuts:

  • Click tray icon - Open settings
  • Right-click tray icon - Quick menu
    • Restart Engine
    • Stop Engine
    • Settings
    • Quit

Troubleshooting Recording:

No transcription appearing:

  1. Check status is "Running" (green)
  2. Verify PTT key is configured
  3. Ensure cursor is in a text field
  4. Try longer recording (>1 second)

Transcription is wrong:

  1. Speak more clearly
  2. Try OpenAI instead of Local
  3. Check microphone quality
  4. Reduce background noise

Sound plays but nothing types:

  1. Click in text field first
  2. Check "Type characters" option
  3. Try different application
  4. Restart engine (right-click tray)

⚙️ Advanced Settings

Environment Variables

DeskTalk respects these env vars:

  • OPENAI_API_KEY - Override API key
  • RUST_LOG=debug - Enable debug logging

Config File Location

%APPDATA%\com.desk-talk.app\config.json

Statistics File

%APPDATA%\com.desk-talk.app\statistics.json

Logs (if enabled)

%APPDATA%\com.desk-talk.app\logs\

🐛 Troubleshooting

Installation Issues

"Windows protected your PC"

  • Click "More info" → "Run anyway"
  • Normal for apps without code signing

Installation fails

  • Run as Administrator
  • Disable antivirus temporarily
  • Check disk space (need ~50 MB)

FFmpeg Issues

"FFmpeg not found"

# Test if FFmpeg is installed
ffmpeg -version

# If not found, install:
winget install Gyan.FFmpeg

# Restart DeskTalk after installing

License Issues

"License validation failed"

"Device already activated"

  • You've activated 3 devices max
  • Deactivate on another PC first
  • Or contact support for device reset

Transcription Issues

"Invalid API key"

  • Get new key from OpenAI
  • Ensure you copied full key (starts with sk-)
  • Check OpenAI account has credits

Transcription too slow

  • Use tiny.en or base.en model
  • Or switch to OpenAI API (faster)
  • Close other heavy applications

Poor accuracy

  • Use OpenAI instead of Local
  • Upgrade to larger local model
  • Improve microphone quality

Audio Issues

No recording

  • Check microphone permissions in Windows
  • Select correct audio device in settings
  • Test mic in Windows Sound settings

Echo or feedback

  • Disable microphone monitoring
  • Use headphones
  • Reduce speaker volume

📊 Pricing

Software License

  • Pro License: $100 (one-time)
  • Includes:
    • Unlimited transcriptions
    • 3 device activations
    • Lifetime updates
    • Email support

OpenAI API Costs (Optional)

  • Whisper API: ~$0.006/minute
  • Pay-as-you-go (no subscription)
  • Example costs:
    • Light user (100 min/month): ~$0.60
    • Medium user (500 min/month): ~$3.00
    • Heavy user (2000 min/month): ~$12.00

Local Model (Free Alternative)

  • $0 forever
  • No per-minute costs
  • No internet required

🔐 Privacy & Security

Data Privacy:

Stored Data:

  • Config: Settings only, no audio
  • Statistics: Word counts, times
  • License: License key (encrypted in Windows Credential Manager)
  • Audio: Temporary files deleted after transcription

Network Usage:

  • License validation: Keygen.sh (on startup/activation)
  • OpenAI API: Only when transcribing (if using cloud)
  • Updates: None (manual updates only for now)

📧 Support

Get Help:

Reporting Bugs:

Include:

  1. Windows version
  2. DeskTalk version (see About)
  3. Error message
  4. Steps to reproduce

🎓 Video Tutorials

Getting Started with DeskTalk: https://youtu.be/SzPE_AE0eEo

How to Get an OpenAI API Key: https://youtu.be/SzPE_AE0eEo

🚀 Updates

Current Version: 0.2.0
Last Updated: October 1, 2025

What's New:

  • Initial public release
  • Keygen licensing system
  • Dark mode support
  • Statistics tracking
  • Error sound feedback

Coming Soon:

  • Auto-updates
  • macOS support
  • Bundled FFmpeg
  • Code signing

📜 License

Copyright © 2025 DeskTalk
See LICENSE file for details

Enjoy faster, hands-free typing with DeskTalk! 🎤✨