System-Level Study on Radar-Based Human Vocal Sound Estimation

This paper explores the use of mmWave radar for estimating human vocal sounds, focusing on its capability to capture vocal fold vibrations during speech. The key contributions include a domain transformation method enabling comparison between radar and acoustic sensors in the same measurement domain, a noise performance analysis demonstrating radar's superiority over tactical throat microphones in noisy environments, and a comparative analysis of radar performance for estimating pitch frequency and harmonics across various radar configurations. The results highlight the potential of mmWave radar as a robust tool for speech sensing, outperforming traditional methods in challenging acoustic conditions while maintaining high accuracy in capturing pitch-related metrics at short distances.