DeepEcho is a comprehensive real-time voice transcription and AI assistant system that supports multiple AI providers. It captures both microphone and speaker audio, provides live transcription, and generates intelligent response suggestions using various AI models including DeepSeek, OpenAI GPT, Claude, Grok, and more.
- 🎤 Real-time Audio Capture: Simultaneous microphone and speaker audio recording
- 📝 Live Transcription: Real-time speech-to-text with local and API modes
- 🤖 Multi-AI Provider Support: DeepSeek, OpenAI, Claude, Grok, Volcano Engine, and GLM
- 🎨 Modern UI: New integrated interface with AI provider selection
- ⚙️ Flexible Configuration: JSON-based configuration with multiple presets
- 🔧 Cross-platform: Windows and macOS support
- 📊 System Monitoring: Built-in diagnostics and resource optimization
- 🛡️ Error Recovery: Comprehensive error handling and retry mechanisms
- Python >=3.8.0
- FFmpeg (for audio processing)
- At least one AI provider API key (see API Setup Guide)
- Windows OS / macOS
-
Clone the repository:
git clone https://github.com/zemochen/deep_echo.git cd deep_echo -
Install dependencies:
pip install -r requirements.txt
-
Set up security (Recommended):
# Linux/macOS chmod +x setup_security.sh ./setup_security.sh # Windows setup_security.bat
This will configure git hooks to prevent accidental API key commits.
-
Set up API keys (choose one method):
Method 1: Configuration File (Recommended)
cp resources/config.example.json config.json # Edit config.json and add your API keyMethod 2: Environment Variable
export DEEPSEEK_API_KEY="sk-your-key-here" # or export OPENAI_API_KEY="sk-your-key-here"
Method 3: Legacy keys.py (Secure)
# Copy the template file cp keys.example.py keys.py # Edit keys.py and add your actual API keys # Note: keys.py is in .gitignore and will NOT be committed to git
⚠️ Security Note: Never commit your actual API keys to version control. Thekeys.pyfile is automatically excluded from git commits. See SECURITY.md for detailed security guidelines. -
Run DeepEcho:
python main.py
python main.py- Uses new integrated architecture
- Automatic AI provider detection
- Modern UI with provider selection
- Comprehensive error handling
python main.py --api- Uses OpenAI Whisper API for transcription
- Higher accuracy and multi-language support
- Requires internet connection
python main.py --legacy- Uses original application architecture
- Backward compatibility mode
- Legacy UI interface
python main.py --verbose- Detailed logging for troubleshooting
- System diagnostics information
| Provider | Models | Setup Guide |
|---|---|---|
| DeepSeek | deepseek-chat, deepseek-coder | DeepSeek Setup |
| OpenAI | gpt-3.5-turbo, gpt-4, gpt-4o | OpenAI Setup |
| Claude | claude-3-haiku, claude-3-sonnet, claude-3-opus | Claude Setup |
| Grok | grok-beta, grok-2 | Grok Setup |
| Volcano Engine | doubao-pro, doubao-lite | Volcano Setup |
| GLM | qwen-turbo, qwen-plus, qwen-max | GLM Setup |
DeepEcho supports multiple configuration presets:
resources/config.example.json- Template configurationresources/config.deepseek.json- DeepSeek optimized settingsresources/config.openai.json- OpenAI optimized settings
{
"audio": {
"use_api_mode": true, // Use API vs local transcription
"record_timeout": 3, // Recording timeout (seconds)
"energy_threshold": 1000 // Audio sensitivity
},
"ai_provider": {
"provider_type": "deepseek", // AI provider to use
"api_key": "your-key-here", // API key
"model": "deepseek-chat", // Model name
"response_interval": 5 // Response update interval
},
"ui": {
"use_new_ui": true, // Use new integrated UI
"theme": "dark", // UI theme
"window_width": 1200 // Window dimensions
}
}- Windows 10/11
- Python 3.8+
- PyAudioWPatch (auto-installed)
- FFmpeg
- macOS 10.14+
- Python 3.8+
- BlackHole virtual audio device
- FFmpeg, PortAudio
macOS Setup:
brew install ffmpeg portaudio python-tk blackhole-2ch- Real-time resource usage tracking
- Thread health monitoring
- Queue size optimization
- Memory usage alerts
- Automatic retry with exponential backoff
- Graceful degradation on failures
- Component health checks
- Network failure handling
- Multi-threaded architecture
- Resource usage optimization
- Queue management
- Memory leak prevention
"FFmpeg not found"
# Windows (with Chocolatey)
choco install ffmpeg
# macOS
brew install ffmpeg"No AI provider configured"
- Check your API key in configuration
- Verify the key format and permissions
- See API Setup Guide
Audio device issues
- Ensure default audio devices are properly set
- On macOS, configure BlackHole for speaker capture
- Check system audio permissions
High memory usage
- Enable resource optimization in config
- Reduce queue size limits
- Use local transcription mode
- Run with verbose logging:
python main.py --verbose - Check the system diagnostics output
- Review configuration file syntax
- Consult the API Setup Guide
- For best accuracy: Use
--apimode with internet connection - For lowest latency: Use local mode with tiny Whisper model
- For cost efficiency: Use DeepSeek or Claude Haiku
- For best quality: Use OpenAI GPT-4 or Claude Opus
If upgrading from an older version:
- Backup your keys.py file
- Create new configuration file:
cp resources/config.example.json config.json
- Update your API keys in the new format
- Test with:
python main.py --legacy(fallback mode)
This project is licensed under the MIT License - see the LICENSE file for details.
Contributions are welcome! Please follow these guidelines:
- Code Style: Follow PEP 8 guidelines
- Testing: Add tests for new features
- Documentation: Update documentation for changes
- Commits: Use clear, descriptive commit messages
pip install -r requirements-dev.txt
pytest tests/- OpenAI for Whisper and GPT models
- Anthropic for Claude models
- DeepSeek for accessible AI APIs
- All contributors and users of DeepEcho