🎙️ Master Transcriber

A powerful Windows desktop application for recording, transcribing, and summarizing meetings and lectures with AI. Built with WPF and C#, featuring real-time transcription, video/audio transcription, multi-language translation, and both local and cloud AI support.

✨ Features

🎯 Core Functionality

• Audio Recording: Capture system audio and microphone simultaneously or separately
• Real-Time Transcription: See transcripts appear live as you speak (configurable 5-300 second intervals)
• Video/Audio Transcription: Transcribe and summarize your video or audio in seconds!
• AI Summarization: Generate concise summaries with key highlights and action items
• Multi-Language Translation: Translate transcripts to 8+ languages using DeepL
• Smart Chunking: Automatically splits long recordings (>10 minutes) for efficient processing
• Markdown Rendering: View AI summaries with beautiful markdown formatting

🔧 Flexible Configuration

• Local or Cloud: Choose between local models (Whisper.cpp, Ollama) or cloud APIs (Azure STT, OpenRouter)
• Mixed Mode: Use local transcription with cloud summarization (or vice versa)
• Customizable Settings: Adjust real-time intervals, enable/disable features, configure API keys

🚀 Quick Start

Prerequisites

• Windows 10/11
• .NET 8.0 Runtime
• (Optional) Whisper.cpp for local transcription
• (Optional) Ollama for local AI summarization

Installation

1. Go to Release page and download the newest version to your PC
2. Extract the contents out from the zip folder, and go into publish folder
3. Double-click TranscribeMeetingUI.exe to run the app
4. Enjoy!

For Dev:

1. Clone the repository

git clone https://github.com/yourusername/meeting-transcriber.git
cd meeting-transcriber

2. Install dependencies

dotnet restore

3. Build the project

dotnet build

4. Run the application

dotnet run --project TranscribeMeetingUI

First-Time Setup

1. Launch the application
2. Click the ⚙️ Settings button
3. Configure your preferred services:
   • For Local Setup: Provide paths to Whisper.cpp and Ollama
   • For Cloud Setup: Enter your API keys (Azure, OpenRouter, DeepL)

1. LLM (Ollama)

The app uses Ollama to run local Large Language Models.

• Installation: Download and install Ollama for your OS.
• Download a Model: Pull a model from the Ollama library. We recommend starting with Llama 3.1 8B.
  Run this command in your terminal:

  ollama pull llama3.1:8b

• Keep Ollama Running: The Ollama background service is all that's needed. The app will automatically detect your downloaded models and show them in the settings panel dropdown.

2. Speech-to-Text (Whisper.cpp)

The app connects to a Whisper.cpp for local transcription.

• Installation: Download the appropriate pre-built binary from the Whisper.cpp releases page.
  • For NVIDIA GPU users on Windows (recommended): Download whisper-cublas-....zip.
  • For CPU-only on Windows: Download whisper-bin-x64.zip.
  • For MAC: Clone the repo and build the APP
• Download a Model: Download a quantized Whisper model. ggml-base.en-q8_0.bin is a fast, high-quality starting point from Huggingface
• Configuration: You will provide the path to whisper-cli.exe and the .bin model file in the app's settings panel. The app will launch and manage the server for you.

---

✅ Cloud Mode (Optional): LLM: OpenRouter STT: Microsoft Azure Speech.

4. Click Save and start recording!

📖 Usage

Basic Recording

1. Toggle Microphone: Enable/disable microphone input (system audio always recorded)
2. Click Start Recording: Begin capturing audio
3. Click Stop Recording: End capture and start processing
4. View Results: See transcript and AI summary
5. Export: Save results in your preferred format

Real-Time Mode

1. Enable Real-Time: Toggle the ⚡ Real-time switch before recording
2. Start Recording: Transcription appears automatically every 30 seconds (configurable)
3. Toggle Translation: Click 🌐 to switch between original and translated text
4. Stop When Done: AI summary generated from all segments

Translation

1. Enable in Settings: Provide DeepL API key and select target language
2. During Recording: Translation happens automatically in real-time mode
3. After Recording: Click 🌐 button to translate on-demand
4. Toggle Views: Switch between original and translated text anytime

⚙️ Configuration

Settings Overview

🎙️ Transcription Provider

• Local (Whisper.cpp): Free, runs on your GPU/CPU
  • Requires: Whisper CLI path and model file (.bin)
  • Models: tiny, base, small, medium, large
• Azure Speech-to-Text: Cloud-based, fast and accurate
  • Requires: Azure API key and region

✨ AI Summary Provider

• Local (Ollama): Free, runs locally
  • Requires: Ollama installed and running
  • Configure: Model name (e.g., llama3.1:8b) and URL
• OpenRouter API: Cloud-based, access to multiple models
  • Requires: OpenRouter API key
  • Configure: Model name (e.g., anthropic/claude-3-sonnet)

🌐 Translation (Optional)

• DeepL API: Professional translation service
  • Requires: DeepL API key (free tier available)
  • Supported: English, Spanish, French, German, Italian, Japanese, Korean, Chinese

⚡ Real-Time Mode

• Chunk Interval: 5-300 seconds (default: 30)
  • Lower = Faster updates, higher CPU usage
  • Higher = Less frequent updates, lower CPU usage

🔧 General

• Enable AI Summary: Toggle summary generation
• Auto-Export: Automatically save after processing

🛠️ Technical Stack

Core Technologies

• Framework: .NET 8.0 / WPF
• Audio: NAudio (WASAPI loopback capture, audio mixing)
• Transcription: Whisper.cpp / Azure Cognitive Services
• AI: Ollama / OpenRouter (Claude, GPT, etc.)
• Translation: DeepL API
• Markdown: Markdig.Wpf for rendering

📁 Project Structure

TranscribeMeetingUI/
├── MainWindow.xaml              # Main UI
├── MainWindow.xaml.cs           # Main logic
├── SettingsWindow.xaml          # Settings UI
├── SettingsWindow.xaml.cs       # Settings logic
├── AppSettings.cs               # Configuration management
├── Recorder.cs                  # Audio recording with mixing
├── RealTimeTranscriber.cs       # Live transcription handler
├── Transcriber.cs               # Local/Azure transcription
├── Summarizer.cs                # Local/Cloud AI summarization
├── ChunkedProcessor.cs          # Long recording handler
├── AzureSTTHandler.cs           # Azure Speech-to-Text
├── OpenRouterHandler.cs         # OpenRouter API integration
├── DeepLTranslator.cs           # DeepL translation

🔒 Privacy & Security

• Local Processing: Use local models (Whisper, Ollama) for complete privacy
• No Data Storage: Application doesn't store recordings or transcripts permanently
• API Keys: Stored locally in %AppData%/MasterTranscriber/settings.json
• Temporary Files: Cleaned up after processing

🐛 Troubleshooting

Common Issues

Whisper.cpp not creating transcript:

• Ensure Whisper CLI path is correct
• Check model file (.bin) exists and is valid
• Verify output directory has write permissions

Audio recording is silent:

• Check system audio is playing
• Verify microphone permissions in Windows settings
• Try toggling microphone on/off

Real-time transcription not working:

• Ensure chunk interval is set (5-300 seconds)
• Check Whisper.cpp or Azure STT is configured
• Look for errors in debug output (Visual Studio Output window)

Translation fails:

• Verify DeepL API key is valid
• Check internet connection
• Ensure target language is supported

Duration mismatch (file longer than actual recording):

• This was a bug in mixing - ensure you're using the latest code
• Check debug output for "Discrepancy ratio" (should be ~1.0x)

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

🙏 Acknowledgments

• NAudio - Audio recording and processing
• Whisper.cpp - Local speech-to-text
• Ollama - Local AI models
• Markdig - Markdown rendering
• Azure Cognitive Services - Cloud transcription
• OpenRouter - Multi-model AI access
• DeepL - Professional translation

---

⭐ If you find this project useful, please consider giving it a star on GitHub!