Jun 25 2019
Nowadays, a number of mobile phones are integrated with a voice-recognition feature, and many times, people are likely to experience an incident where a speech recognition application is triggered in the middle of a conversation or a meeting in the office.
The research group led by Professor Kilwon Cho and Professor Yoonyoung Chung developed a skin-attachable vibration sensor for voice recognition. (Image credit: Pohang University of Science & Technology)
At times, it is not at all activated, in spite of calling out the application a number of times. The reason for this is that a mobile phone utilizes a microphone, which perceives sound pressure to detect voice and, moreover, it is easily influenced by various obstacles, including the surrounding noise.
In this context, a wearable and flexible vibration responsive sensor was successfully created by Professor Kilwon Cho of Chemical Engineering and also by Professor Yoonyoung Chung of Electronic and Electric Engineering, both from the Pohang University of Science & Technology (POSTECH).
When fixed to a person’s neck, the novel sensor can accurately detect voice via vibration of the neck skin, but it is no way influenced by the sound volume or ambient noise.
The traditional vibration sensors can perceive voice via air vibration, but their sensitivity reduces because of the damping effect and mechanical resonance; as a result, these sensors do not have the potential to measure voices in a quantitative manner. Therefore, obstacles or ambient sound, like a mouth mask, can have an impact on its precision of voice recognition and hence, it is not suitable for use in security authentication.
In this analysis, the researchers showed that the voice pressure is in proportion to the acceleration of the vibration of the neck skin at different levels of sound pressure ranging from 40 to 70 dB SPL, and they successfully created a vibration sensor using the acceleration of skin vibration.
The device, containing a diaphragm with very small holes and an ultrathin polymer film, can quantitatively perceive voices by determining the acceleration of skin vibration.
In addition, the researchers effectively demonstrated that even at a very volume of low voice with a mouth mask worn and also in a noisy setting, the novel device can precisely sense voice without vibrational distortion.
The study could also be applied to a wide range of voice-recognition applications, for example, a wearable vocal healthcare monitoring device, human-machine interface, and electronic skin.
This research is very meaningful in a way that it developed a new voice-recognition system which can quantitatively sense and analyze voice and is not affected by the surroundings. It took a step forward from the conventional voice-recognition system that could only recognize voice qualitatively.
Kilwon Cho, Professor, Department of Chemical Engineering, POSTECH
Thus, Professor Cho described the meaning of the study during an interview.
The Center for Advanced Soft Electronics under the Global Frontier Research Program of The Ministry of Science and ICT, Korea, supported the study. Additional results of this study can be found on the website of Nature Communications, published on June 18th, 2019.