WhisperMate

A native macOS voice-to-text app powered by Groq's Whisper API.

Key Features

Security & Privacy

• Open source - read the code to see exactly how data is handled
• Audio is transcribed and immediately discarded, nothing is stored
• Data only sent to Groq API, no third-party servers
• API keys stored in macOS Keychain

Performance

• 400-800ms response times using Groq's inference
• Native Swift/SwiftUI implementation
• 1.35 MB app size (vs 200+ MB for Electron-based alternatives)
• Minimal CPU and memory usage

LLM Post-Processing

• Optional transformations using Groq's LLM models
• Translate between languages (e.g. speak Russian, get English text)
• Adjust tone or formality
• Custom glossaries for domain-specific terms

Usage

• Press and hold a hotkey (like Fn) to record
• Double-tap for continuous recording mode
• Auto-paste transcriptions into active application
• Minimal overlay indicator or full window mode

Pricing

Free during beta. You only pay for Groq API usage.

Installation

Option 1: Download Release (Recommended)

1. Download the latest WhisperMate-v0.0.4.dmg from the Releases page
2. Open the DMG file
3. Drag Whispermate to your Applications folder
4. Launch Whispermate from Applications
5. Follow the onboarding wizard to:
   • Grant microphone permission
   • Grant accessibility permission (for auto-paste)
   • Configure your recording hotkey

Option 2: Build from Source

Prerequisites

• macOS 13.0+ (Ventura or later)
• Xcode 15.0+
• Groq API key (get one at https://console.groq.com)

Steps

1. Clone the repository:

   git clone https://github.com/writingmate/whispermate.git
   cd whispermate/Whishpermate

2. Open the project:

   open Whispermate.xcodeproj

3. Build and run (⌘R)

4. Follow the onboarding wizard on first launch

Getting Started

1. First Launch: Complete the onboarding wizard

   • Enable microphone access
   • Enable accessibility permissions (needed for auto-paste)
   • Set your recording hotkey (Fn key recommended)

2. Recording:

   • Hold-to-Record: Press and hold your hotkey, release to transcribe
   • Continuous Recording: Double-tap your hotkey to start, tap once to stop

3. Modes:

   • Overlay Mode: Minimal indicator in bottom-right corner
   • Full Mode: Expanded window with settings and transcription history

4. Settings:

   • Configure your Groq API key
   • Choose between OpenAI Whisper or Groq transcription
   • Customize hotkey
   • Toggle auto-paste functionality

Security

• API keys are stored securely in macOS Keychain (never in plain text)
• Audio files are temporary and not persisted
• No data is sent anywhere except to Groq's API

Architecture

WhisperMate/
├── WhisperMateApp.swift          # App entry point
├── Views/
│   └── ContentView.swift          # Main UI
├── Services/
│   ├── AudioRecorder.swift        # Audio recording with AVFoundation
│   ├── GroqAPIClient.swift        # Groq API integration
│   └── KeychainHelper.swift       # Secure key storage
└── Info.plist                     # App permissions

Groq API

This app uses Groq's Whisper Large V3 model for transcription:

• Fast inference times (typically < 1 second)
• High accuracy
• Cost-effective

Get your API key: https://console.groq.com

License

MIT