Audio Transcription App v2.0.0

A powerful desktop application for audio transcription and AI-powered analysis. Transcribe audio files using OpenAI's advanced models, then chat with your transcripts using intelligent AI assistance.

✨ Key Features

🎙️ Recording & Transcription

• Multiple transcription models - Choose from gpt-4o-transcribe, whisper-1, or gpt-4o-transcribe-diarize
• Speaker diarization - Automatic speaker identification with reference audio support
• Direct audio recording - Record audio directly in the app with live waveform visualization
• Drag-and-drop upload - Support for MP3, WAV, M4A, WEBM, MP4, OGG, FLAC formats
• Large file support - Automatic chunking for files of any size (no 25MB limit)
• Custom prompts - Guide transcription with context-specific prompts
• Summary generation - AI-powered summaries with customizable templates

🚀 Performance Optimizations (NEW in v2.0.0)

• Parallel chunk processing - 60-80% faster transcription for large files
• Dynamic rate limiting - Intelligent API request management (80 RPM)
• Audio speed optimization - Optional 2-3x speed-up for 23-33% cost savings
• Opus compression - Bandwidth optimization with 5-10x file size reduction

💬 AI-Powered Analysis

• Intelligent chat interface - Ask questions about your transcripts using OpenAI Agents SDK
• Context-aware responses - AI fetches only relevant sections (90% token reduction)
• Multi-transcript support - Compare and analyze multiple transcripts simultaneously
• Advanced tools - Search, chunk retrieval, speaker extraction, transcript comparison
• Chat history - Persistent conversation history per transcript

📚 Transcript Library

• Organized storage - All transcripts saved automatically with metadata
• Smart search - Find transcripts by name, content, or date
• Filtering options - View All, Starred, or Recent transcripts
• Export formats - TXT, VTT, or Markdown with one click
• Secure storage - API keys stored in system keychain (macOS/Windows)

🎨 Modern Interface

• Two-tab design - Separate Recording and Analysis workspaces
• Dark mode - Beautiful dark theme with system-aware switching
• Resizable panels - Customize your workspace layout
• Apple-inspired design - Clean, minimal, and intuitive

📦 Installation

macOS

Download and install Audio Transcription-2.0.0-arm64.dmg

Windows

Download and run Audio Transcription Setup 2.0.0.exe or use the portable version

🚀 Quick Start

1. Launch the app and enter your OpenAI API key
2. Recording tab: Upload or record audio, then transcribe
3. Analysis tab: View transcripts and chat with AI about the content

📖 Usage Guide

Recording Tab

1. Upload or Record Audio

   • Drag and drop an audio file
   • Click "Choose File" to browse
   • Or use "Record" to capture audio directly

2. Configure Transcription

   • Select transcription model (gpt-4o-transcribe recommended)
   • Enable speaker diarization if needed
   • Add optional context prompt
   • Choose summary template

3. Transcribe

   • Click "Transcribe" and monitor progress
   • Large files automatically chunked and processed in parallel
   • Transcript auto-saves to Analysis tab when complete

Analysis Tab

1. Transcript Library (Left Panel)

   • Search transcripts by name
   • Filter: All / Starred / Recent
   • Click to view transcript

2. Transcript Viewer (Middle Panel)

   • Read full transcript with formatting
   • Export to TXT, VTT, or Markdown
   • Star important transcripts

3. AI Chat (Right Panel)

   • Select one or more transcripts for context
   • Ask questions about the content
   • AI intelligently searches and references specific sections
   • Chat history saved per transcript

Example Chat Queries

• "What were the main topics discussed?"
• "Summarize the key decisions made"
• "What did [Speaker Name] say about [topic]?"
• "Find all mentions of [keyword]"
• "Compare how the speakers approached [topic]"

🔧 Technical Details

Architecture

• Frontend: React 19, Vite, TailwindCSS
• Backend: Electron 28, Node.js
• AI: OpenAI Agents SDK, gpt-4o, Whisper models
• Storage: electron-store with system keychain integration
• Audio: FFmpeg with fluent-ffmpeg wrapper

Transcription Models

• gpt-4o-transcribe - Latest model, best quality, $0.006/minute
• whisper-1 - Previous generation, $0.006/minute
• gpt-4o-transcribe-diarize - Automatic speaker identification

Performance Features

• Parallel chunk processing (5 concurrent)
• Dynamic rate limiting (80 RPM)
• Optional audio speed optimization (1x-3x)
• Optional Opus compression for uploads
• Automatic format conversion (OGG, FLAC → MP3)

Security & Privacy

• API keys stored in system keychain (macOS Keychain/Windows Credential Manager)
• All data stored locally (no cloud sync)
• Chat history encrypted with OS-level encryption

🔑 API Key Setup

1. Get your API key from OpenAI Platform
2. Click the key icon in the app header
3. Paste your API key and click "Save"
4. Key is securely stored in your system keychain

🧪 Testing

Prerequisites

• Node.js 20+
• OpenAI API key

Setup

# Install dependencies
npm install

# Run in development mode
npm start

# Run tests
export OPENAI_API_KEY=your-api-key-here
npm test

# Build for macOS
npm run build:mac

# Build for Windows
npm run build:win

Test Files

• test-ffmpeg.js - FFmpeg infrastructure tests
• test-transcription-service.js - Integration tests for optimizations

📊 Performance Benchmarks

Large File Example (60 min audio, 10 chunks)

v1.0.0 (Sequential):

• Processing time: ~320 seconds
• Cost: $0.36

v2.0.0 (Parallel):

• Processing time: ~70 seconds (78% faster)
• Cost: $0.36

v2.0.0 (Parallel + 2.5x Speed):

• Processing time: ~50 seconds
• Cost: $0.27 (25% savings)

🛠️ Development

See CLAUDE.md for comprehensive development documentation including:

• Project architecture
• Backend services structure
• Adding new features
• Agent tools and guardrails
• Testing strategies

📝 Credits

Created by Patrick C. Freyer and Alexander Achba

Open Source Libraries

• Electron
• React
• OpenAI Node SDK
• OpenAI Agents SDK
• FFmpeg

📄 License

MIT

🔗 Links

• Repository
• OpenAI Platform
• Issue Tracker

---

Version 2.0.0 - Major redesign with AI chat, performance optimizations, and comprehensive analysis features