Physical Formant Speech Synthesis

Sonon embeds a pure mathematical articulatory speech synthesizer built on:

• Liljencrants-Fant Glottal Flow: Parametric differential model generating natural vocal cord excitation pulses with adjustable open quotient ($O_q$), return phase ($R_a$), and spectral tilt ($R_k$).
• 4-Pole Cascade Vocal Tract Filter: Second-order resonator poles implementing resonant formants $F_1, F_2, F_3, F_4$.

This allows generating synthesized command words directly on-device without neural network memory overhead.

Try synthesizing phrases in the Playground Speech Synthesis Tab.