Fig. 3: Comparative experiments on speech detection in harsh acoustic environments.
From: Machine learning-assisted wearable sensing systems for speech recognition and interaction
a Photographs of the same comparison test in a quiet environment (I), a noisy environment (II), and wearing a mask (III). b When the participant said “CQU” in a quiet environment (I), a noisy environment (II), and wearing a mask (III), SAAS showed the time domain waveform and spectrum information of the sound signal. c When the participant said “CQU” in quiet environments (I), noisy environments (II), and wearing masks (III), the commercial reference microphone showed the time domain waveform and spectrum information of the sound signal. d Photos of the test subjects in 9 throat attachment positions and movements. e Time domain waveforms obtained when saying “Perfect” through SAAS in 9 throat attachment positions and movements.
