An open-source modular voice assistant and general ambient audio processor. Speech recognition (via DeepSpeech) and keyword detection (via porcupine) are all done on device to avoid network communication of audio or speech transcriptions. Ambient audio processing can also be done, eventually for visualization using AdaFruit NeoPixel. This is meant to run on a Raspberry Pi.

• Keyword detection and speech recognition
• Simple phrase based queries
• TTS responses
• Spectral flux based onset detection

• NeoPixel integration for ambient/music visualization
• Beat visualization using onset detection
• Other types of visualization modes
• Other types of queries
• Integration with streaming services like Spotify
• Integration with YAMNet for context-aware queries (and for fun visualizations)

This is a work in progress!