Foreign researchers have developed wearable vibration sensors for speech recognition

As an emerging interdisciplinary field, speech recognition technology has made significant progress over the past two decades and has already entered the commercial market. Someone once predicted that within 10 years, speech recognition technology would enter various fields of daily life, such as home appliances, communications, healthcare, and home services. Especially now, speech recognition is more integrated into our lives, and their presence can be seen in widely popular smartphones.

However, due to the presence of obstacles and other noise in the surrounding environment, the performance of speech recognition is always somewhat unstable, and sometimes even misidentification may occur. This is because microphones are commonly used in mobile phones to detect sound. A microphone, in simple terms, is an energy conversion device that converts sound signals into electrical signals. The initial microphones were converted from sound to electricity through resistors, and later developed into sound to electricity conversion through inductors or capacitors. Because they work by sensing vibrations in the air, they all have a common drawback of being susceptible to environmental noise.

According to the Sensor Expert Network, a recent foreign research achievement has successfully solved this problem. Researchers at Pohang University of Science and Technology (POSTECH) have successfully developed a flexible and wearable vibration response sensor that can accurately recognize the user's voice. When connected to the neck, the sensor can recognize sound through the vibration of the neck skin and is not affected by environmental noise or sound volume.

These skin attached sensors detect and recognize human voice by examining neck skin vibration, which can be measured through several parameters including velocity, displacement, and acceleration.

Traditional vibration sensors recognize sound through air vibration and have reduced sensitivity due to mechanical resonance and damping effects. Therefore, they are unable to quantitatively measure sound. Therefore, environmental sounds or obstacles such as masks can affect the accuracy of speech recognition and cannot be used for security authentication.

In this study, the research team demonstrated that sound pressure is directly proportional to neck skin vibration acceleration at various sound pressure levels, ranging from 40 to 70 dB SPL (sound pressure level). Researchers have developed a new wearable vibration sensor that utilizes skin vibration acceleration. The device consists of an ultra-thin polymer film and a diaphragm with small holes. It can quantitatively perceive sound by measuring the acceleration of skin vibration. In a successful demonstration, wearable sensors were able to accurately recognize speech even in noisy environments without vibration distortion, and the volume was very low when wearing a mask.

This research can be further extended to various speech recognition applications, such as electronic skins, human-machine interfaces, and wearable vocal health monitoring devices.

The researchers explained the significance of this study in his interview. This research is very meaningful as it develops a new speech recognition system that can quantitatively perceive and analyze speech without being affected by the surrounding environment. It takes a step forward from traditional speech recognition systems, which can only recognize speech qualitatively. This research is supported by the Advanced Soft Electronics Center under the Frontier Research Program of the South Korean Ministry of Science and Information and Communication Technology.