Use TTS to create synthetic audio data for hotword detection
Poor mans way to get audio samples for a specific hotword. This uses different free linux TTS systems to generate a range of audio samples for hotword detection.
Don't expect this to work very well in production, as due to the limited number of voices and low variety this tends to overfit quickly