This type furnishes rules that govern how the synthesized voice should be produced, including spectrogram generation.