MemoMaker: Audio to AI Memos

By

Transform audio recordings into professional transcripts and actionable AI memos using Google’s Gemini. AI memo generator for business now has a built-in recording capability!

audio to ai memos

Python License AI

✨ AI Memos Key Features

  • 🎤 Built-in Recording – Record meetings directly with one-click start/stop
  • 🎯 Smart Audio Processing – Automatic detection of optimal processing method
  • 📱 Speech-Optimized – 22kHz mono recording perfect for meetings
  • 💾 Efficient MP3 Encoding – Small file sizes without external dependencies
  • 📁 Organized Storage – Auto-creates recordings folder with timestamped files
  • 🎨 Modern UI – Clean dark-themed interface with real-time progress tracking
  • 🌍 AI Memos Multi-Language Support – Estonian and English prompts with easy language switching
  • ⚡ Multiple Processing Methods – Inline, cloud upload, or auto-detection
  • 📝 Configurable Prompts – Customize transcription and generate AI memos
  • 🔊 Wide Audio Support – MP3, WAV, M4A, OGG, FLAC, AAC formats
  • 📋 Markdown Output – Professional memo format with timestamps and action items
  • 🛡️ File Validation – Comprehensive format, size, and integrity checking
  • 📊 API Usage Tracking – Real-time token counts and processing statistics
  • 🔄 Real Progress Bar – Step-by-step progress indication during processing
  • 🔑 Built-in API Key Manager – Easy setup and management of Gemini API keys

🚀 Prerequisites

  1. Python 3.8+
  2. Google API Key – Get one for free from Google AI Studio
  3. Required packages:
    pip install customtkinter google-generativeai pillow sounddevice scipy numpy lameenc

How to use AI memo writer: Setup AI Memos

  1. Clone the repository:
    git clone https://github.com/priit2000/memomaker.git
cd memomaker
  2. Set your Gemini API key:
    export GEMINI_API_KEY="your-api-key-here"

    Or on Windows:

    set GEMINI_API_KEY=your-api-key-here

    Or use the built-in API key manager: The app will show a setup dialog if no API key is found.

  3. Run the application:
    python memomaker-ui.py

🎯 How to Use AI memos

Recording Mode (New!)

  1. Launch the app – Run python memomaker-ui.py
  2. Start recording – Click “🎤 Start Recording” button
  3. Record your meeting – Speak clearly into your microphone
  4. Stop recording – Click “🛑 Stop Recording” when finished
  5. Auto-processing – App automatically processes the recording and generates transcript + memo
  6. View results – Files saved in recordings/ folder with timestamp naming

File Processing Mode

  1. Launch the app – Run python memomaker-ui.py
  2. Select audio file – Click “Browse” and choose your audio file (or click the file path field)
  3. Choose language – Select Estonian (ET) or English (EN) from the language dropdown
  4. Choose processing method:
    • 🎯 Auto – Smart detection based on file size
    • ⚡ Inline – Fast processing for smaller files (<20MB)
    • ☁️ Cloud Upload – Better for larger files (>20MB)
  5. Customize prompts (optional) – Edit transcription and memo prompts in the tabs
  6. Process – Click “Process Audio” and watch real-time progress
  7. Manage API key (optional) – Click “🔑 API Key” button to view/edit your Gemini API key
  8. View results – Files saved in recordings/ folder with organized naming
  9. Monitor usage – View detailed API usage statistics including token counts in the results area

CLI Mode

python memomaker-ui.py audio_file.mp3 [--method auto|inline|upload] [--prompt "custom prompt"]

📁 File Structure

memomaker/
├── memomaker-ui.py              # Main application
├── transcription-prompt-et.md   # Estonian prompts
├── transcription-prompt-en.md   # English prompts
├── .gitignore                  # Git ignore rules
├── README.md                   # This file
└── recordings/                 # Auto-created folder for all outputs
    ├── 241113-143022-recording.mp3    # Recorded audio
    ├── 241113-143022-transcript.txt   # Generated transcript
    └── 241113-143022-memo.md          # Generated memo

📊 Key Features Detailed

🎤 Built-in Audio Recording

  • One-click recording: Start/stop with visual feedback
  • Speech-optimized: 22kHz mono recording perfect for meetings
  • Efficient encoding: Direct MP3 encoding with lameenc (no external tools needed)
  • Auto-processing: Automatically processes recorded audio when recording stops
  • File size optimization: ~1MB per minute vs ~10MB for standard recording

📁 Organized File Management

  • Auto-folder creation: recordings/ folder created automatically
  • Timestamped naming: yymmdd-hhmmss-[type] format for easy organization
  • Session grouping: All files from same recording session have matching timestamps
  • Example: 241113-143022-recording.mp3, 241113-143022-transcript.txt, 241113-143022-memo.md

⚙️ Configuration

Settings (Top of memomaker-ui.py)

# API Configuration
API_KEY = os.environ.get("GEMINI_API_KEY")

# Model Settings
MODEL_NAME = 'gemini-flash-latest'

# File Processing Settings
INLINE_THRESHOLD = 20 * 1024 * 1024  # 20 MB
MAX_FILE_SIZE = 100 * 1024 * 1024     # 100 MB max
MIN_FILE_SIZE = 1024                   # 1 KB min

# UI Settings
WINDOW_WIDTH = 1000
WINDOW_HEIGHT = 800

Multi-Language AI Memo Prompts

The app automatically detects and uses language-specific prompt files:

  • Estonian: transcription-prompt-et.md
  • English: transcription-prompt-en.md
  • Future: transcription-prompt-fr.md, transcription-prompt-de.md, etc.

Each file contains:

  • Transcription rules – Under # Transkriptsioon/# Transcription section
  • Memo format – Under # Memo section

📝 AI Memos Output Examples

Transcript Format

[00h:02m:15s] Priit Kallas: Alustame tänase koosoleku. Päevakorras on kolm punkti.
[00h:02m:28s] Henrik Aavik: Tänan. Kas võiksime alustada eelmise nädala tulemustega?
[00h:02m:45s] Priit Kallas: Kindlasti. Numbrid on väga head...

AI Memo Format (Markdown)

  • Structured sections: Participants, summary, decisions, actions
  • Timestamps: References to specific moments in audio
  • Action items: Clear responsibilities and deadlines
  • Multi-language: Professional business language (Estonian or English)
  • Markdown format: Easy to edit and convert to other formats

🔧 Troubleshooting

Common Issues

“No module named ‘customtkinter'”

pip install customtkinter sounddevice scipy numpy lameenc

“Invalid API key”

  • Verify your Google API key is correct
  • Check environment variable is set: echo $GEMINI_API_KEY (Linux/Mac) or echo %GEMINI_API_KEY% (Windows)
  • Use the built-in “🔑 API Key” button to set up your key

“File validation failed” errors

  • Check file format (supported: MP3, WAV, M4A, OGG, FLAC, AAC)
  • Ensure file size is between 1KB and 100MB
  • Use “Upload” method for files > 20MB
  • Consider compressing large audio files

“Recording not working”

  • Check microphone permissions in your system settings
  • Install audio dependencies: pip install sounddevice scipy numpy lameenc
  • Test microphone with other applications first

Performance Tips

  • Use built-in recording for optimal file sizes and quality
  • Recording produces ~1MB per minute vs ~10MB for standard recording
  • Compress large files before processing
  • You can change the Gemini model in the variables section of the py file

📋 Roadmap

  • Built-in recording – Direct audio capture (completed)
  • MP3 optimization – Speech-optimized recording (completed)
  • Organized file structure – Timestamped file naming (completed)
  • Batch processing – Process multiple files
  • Export formats – PDF, Word, plain text
  • Audio player – Built-in playback with waveform
  • Cloud storage – Direct integration with Google Drive/OneDrive
  • Multi-language – Estonian and English support (completed)
  • Additional languages – French, German, etc.
  • Templates – Custom memo templates

⚠️ Security & Privacy

  • API keys are stored locally as environment variables
  • Audio files are processed according to Google’s privacy policy
  • No data retention – Files are not stored after processing
  • Local processing – Transcripts and memos saved locally
  • Recordings stay local – All recorded audio stored in local recordings folder

📄 License

MIT License – see LICENSE file for details.

📊 API Usage Tracking

  • Real-time statistics displayed in results area
  • Token counts: Input, output, and total tokens
  • Processing time: Detailed timing for each operation

File Validation

  • Format checking: Validates audio file extensions and MIME types
  • Size limits: Enforces minimum (1KB) and maximum (100MB) file sizes
  • Integrity checks: Basic corruption detection
  • Clear error messages: Specific validation failure details

🙏 Acknowledgments

  • Google Gemini AI – For powerful audio processing capabilities
  • CustomTkinter – For modern UI components
  • LAME MP3 Encoder – For efficient audio compression

Related Posts

Amperly AI Agentuur

Privacy Policy