FRONT-END SYSTEM STRUCTURE
Each algorithm converts speech audio data into observation vectors.
