Electronics engineers have created a wide range of systems based on human and animal biology in recent years. By intently duplicating complex natural cycles, these frameworks could at last handle various genuine issues all the more reliably and effectively.
A new microelectromechanical sensor was recently developed by researchers at Technische Universität Ilmenau, Kiel University, University College Cork, Karlsruher Institute of Technology, and Fraunhofer Institute for Digital Media Technology Ilmenau. The cochlea is a spiral-shaped cavity in the inner ear of the human ear that produces nerve impulses in response to vibrations of sounds. In contrast to simple microphones, this artificial cochlea, which was presented in Nature Electronics, is better able to process sounds and identify individual signals in noisy environments.
“The thought for the examination was to assemble counterfeit hair cells, like the hair cells in the inward ear, which are answerable for the sound recognition in the human ear,” Claudia Lenk, lead creator of the paper, told Tech Xplore. “We were inspired by Bryan Joyce et al.’s previous work, which used cantilevers with complex feedback to get the hair cells’ specific properties. Because we have a specific kind of cantilever (silicon beams) that can be read out and actuated completely electronically, we thought we could make artificial hair cells quickly and easily.”
“The goal of the research was to create artificial hair cells that functioned similarly to the hair cells in the inner ear, which are responsible for sound detection in the human ear,”
Claudia Lenk, lead author of the paper,
The human ear and the ears of different vertebrates work uniquely in contrast to traditional mouthpieces. As receivers distinguish all tones (i.e., frequencies) of a sound sign all the time, the noisy tones they get can, in some cases, cover calmer tones. On the other hand, distinct tones (sound signals in various frequency bands) are detected by distinct hair cells in the ears.
One MEMS cochlea sensor (cantilever) (without feedback). Credit: TU Ilmenau/Michael Reichel
“Hair cells in the ears can be individually tuned, especially the gain for detection, and so how each tone is detected can be adjusted,” Lenk explained. This is significant to isolate, e.g., discourse from foundational signals. The gain would be high for the speech signal’s tones and low for the background signals in this scenario. Background and speech can thus be separated much more easily.”
There are two essential parts to the microelectromechanical cochlea that Lenk and her colleagues developed. The first consists of a series of tiny silicon cantilevers that are made to look like real hair cells and detect sound signals from the environment. The second is a feedback loop that individually adjusts each cantilever’s detection capabilities.
Lenk stated, “If sound is impinging, the cantilevers oscillate, and this oscillation is measured as an electronic signal by an integrated readout system.” An actuator is also present in each cantilever. The cantilever is compelled to bend or oscillate, depending on the applied signal, when a voltage is applied to the actuator. For the loop of feedback, an offset (static) voltage is added, and the measured signal is amplified. This signal controls the cantilever’s actuator.
The input signal delivered by the counterfeit cochlea’s criticism circle eventually changes how sound is distinguished by individual cantilevers, for example, by regulating its benefit, awareness, and transmission capacity. The cantilever’s gain can also be set to be nonlinear, allowing it to change in response to the amplitude of the sounds it picks up. This means that quieter sounds can be made louder, and louder sounds can be made quieter. This makes sure that a human listener can hear all recorded sounds.
Comparative evaluation of the bio-inspired (smart) sensor with a conventional microphone. Credit: TU Ilmenau/Michael Reichel
“Since each cantilever reacts only to a specific frequency range of the sound,” Lenk stated, “an array of cantilevers can be used to decompose the signal into the responses for different frequencies, similar to how it is done in the human ear.” Additionally, since each cantilever’s detection properties can be individually adjusted, the manner in which various tones are sensed can be altered. The system is extremely adaptable due to its combination of tone-specific response, nonlinear response, and feedback-tuning response.
The high flexibility of the microelectromechanical cochlea presented by Lenk and her partners could be very favorable for genuine applications, as various conditions (e.g., a peaceful room, a train station, a shopping center, and so on.) can have changing measures of foundation commotion. Their system might be able to pick up a lot more sounds than conventional microphones, which have fixed properties, if they could change how the cantilevers sense sound.
If the sounds were recorded in a crowded and noisy environment, microphone recordings must be thoroughly analyzed with specialized software to identify some of the quieter signals. On the other hand, the researchers’ adaptive cochlea has the ability to apply changes and filter out sounds earlier (i.e., while it is recording signals). This can make a number of real-world applications, like speech recognition, simpler, making these systems better able to classify sounds that are collected in noisy environments.
According to Lenk, “We successfully implemented the properties of human hearing directly in the sensor via a quite simple and quick feedback loop.” This has two major ramifications. First and foremost, the framework is quickly versatile, and this is very significant for application in different circumstances. The bio-inspired properties of the cochlea also help to highlight important information like when a sound started or specific tones, which should help to speed up and make processing (like speech detection) faster and more efficient (“smaller neural networks are needed for processing”).
Comparative evaluation of the bio-inspired (smart) sensor with a conventional microphone. Credit: TU Ilmenau/Michael Reichel
Another advantage of the team’s artificial cochlea is its ease of integration into a broader system. It could be used to improve smart speakers, security systems, hearing aids, and other products.
Lenk explained, “Other approaches to building such bio-inspired properties are more complex.” Because of this, it is more challenging for them to scale to the entire hearing range for use in devices like hearing aids, smart speakers, and so on. and they rarely encourage adaptation.”
Hearing misfortune is regularly brought about by a deficiency of hair cells inside the inward ear, which thus restricts the capacity to hear explicit sounds. The team is currently evaluating the possibility that Lenk and her colleagues’ new artificial cochlea, which replicates the function of hair cells, could be used to develop better hearing aids.
Lenk went on to say, “Our next steps will be to build a system that covers the whole range of hearing and combine the MEMS cochlea with neural networks for processing.” The last option is essential to empower programmed transformation and underscore significant sounds, similar to discourse. This framework will be produced for the utilization of amplifiers in our new task, NeuroSensEar.”
More information: Claudia Lenk et al, Neuromorphic acoustic sensing using an adaptive microelectromechanical cochlea with integrated feedback, Nature Electronics (2023). DOI: 10.1038/s41928-023-00957-5
