KR102126204B1 - Voice Recognition Sensor having Multi Frequency Channels with Curved type - Google Patents

Abstract

The voice recognition sensor according to the present invention includes a plurality of frequency channels having different lengths, the plurality of frequency channels senses a frequency region overlapping each other, and the overall shape of continuously connecting the outsides of the plurality of frequency channels is a curve. It forms a curved surface, and through this, the plurality of frequency channels is characterized by having a response characteristic having high sensitivity while simultaneously sensing frequencies through an audible frequency range of 200 Hz to 4 kHz.

Description

Voice Recognition Sensor having Multi Frequency Channels with Curved type

The present invention relates to a low-power flexible piezoelectric voice recognition sensor having a plurality of curved frequency channels for Internet of Thing (IoT) applications, and more specifically, it is capable of sensitively sensing a person's voice band as a whole. The present invention relates to a flexible piezoelectric-based low-power voice recognition sensor that enables multiple frequency bands to be uniformly set as a whole by arranging a plurality of frequency channels in a curved form.

The speech recognition sensor refers to a sensor that extracts linguistic information from acoustic information contained in human speech and makes it perceive and react. Today, voice communication is considered to be the most natural and convenient way to exchange information between human and machine information in the future IoT era. However, in order to communicate with the machine through speech, human speech must be converted into a format that can be processed by the machine. This process is speech recognition.

Voice recognition, represented by Apple's Siri, is composed of a combination of a microphone, analog to digital converter (ADC), and digital signal processing (DSP). It is operated by pressing the end button. This is one of the biggest challenges in realizing voice recognition-based Internet of Things (Internet of Things), and it is expected to open endless IoT applications when developing a low-power, always-on voice recognition system.

The voice recognition system, which can be easily used without learning or training, is a promising technology that will lead the future industry in the IoT era where the demand for UI development and construction for innovative next-generation IT products is high. Since it can be input and the input speed is faster than typing, it has the advantage of being able to process information in high speed or in real time.

In recent years, the evolution of smartphone terminal performance, the development of artificial intelligence and knowledge retrieval technology, and the processing of large amounts of data through cloud-based voice recognition systems are intelligent agents that enable users to find the answers they want accurately and quickly. Despite the possibilities, the speech recognition technology still has the following limitations.

First, from a hardware point of view, the existing voice recognition technology using a combination of microphone, ADC, and DSP has a very high power consumption, so voice recognition in a standby mode without a separate power source is practically impossible. Moreover, it is applied to voice recognition sensors for mobile applications. Silver is very limited due to energy problems. In addition, preliminary actions, such as pressing a voice recognition start button, are required, and the accuracy, reliability, and speed are deteriorated. In other words, high sensitivity is essential to apply to IoT-based smartphones, TVs, automobiles, and other wearable devices, and it is necessary to be able to recognize the user's voice with ultra-power by maintaining a standby state without excessive power consumption even in sleep state. do.

Next, from the perspective of acoustics and linguistics, the speech recognition of current microphone, ADC, and DSP combinations is based on complex algorithms, which limits the recognition of natural conversation.

On the other hand, human cochlea is efficiently processing complex language after frequency separation through simple algorithm. Despite the various devices using the principle of the cochlear, there is a precedent for simulating it and applying it to a cochlear implant, but the case used as a low-power voice recognition sensor for IoT has not yet been used.

Additionally, the existing voice sensor may be generally composed of a sensor unit and readout integrated circuits (ROIC), and a cap type voice sensor should always provide a bias on the sensor unit.

Flexible piezoelectric thin film cochlear implant applications are described in H. Lee et. al's Journal of the Advanced Functional Materials Journal Vol. 24, No. 44, pg 6914, 2014. Three piezoelectric elements were attached on a thin trapezoidal silicon membrane to separate the audio signals in the audible frequency band. In the above document, three individual piezoelectric elements are attached to the silicon membrane to separate the frequencies and applied to the cochlear implant, but the algorithm and circuit design are not considered as low-power voice sensors for IoT.

On the other hand, even in the case of the pre-registered patent No. 10-1718214 by the present applicant, the separated voice signal is piezoelectrically separated by separating the voice sensed through the plurality of channels according to the frequency using a plurality of frequency separation channels formed in a trapezoidal shape Although it discloses the technical content to recognize by converting the mechanical vibration signal to an electrical signal through the device, there is a limitation that response performance sensed through a plurality of frequency channels cannot be smoothly implemented in a certain area.

Also, as a conventional document presenting a piezoelectric device that outputs a haptic feedback effect using a plurality of resonant frequencies, reference may be made to Patent Publication No. 10-2012-0099036 (2012.09.06). On the other hand, the above document provides a haptic feedback technology based on tactile sense, force, and movement, but there is a limitation that it does not disclose how to recognize the recognized voice in a state separated into a plurality of frequencies.

(Thesis) H. Lee et. al, Advanced Functional Materials, 24(44), 6914, 2014

(Patent Document 1) KR10-1718214 B

(Patent Document 2) KR10-2012-0099036 A

The present invention is intended to solve the above-described problems, and separates voices sensed through a plurality of frequency separation channels formed in a curved form using a flexible piezoelectric thin film made of a single element through the plurality of channels according to frequency. At the same time, it is possible to provide a low-power piezoelectric voice recognition sensor for IoT that reduces power consumption through simplification of the voice recognition circuit by recognizing the separated voice signal by converting it from a mechanical vibration signal to an electrical signal through a flexible inorganic piezoelectric element. Purpose.

That is, in a conventional structure in which a plurality of frequency-separated channels are arranged in a trapezoidal shape, a problem in which a sensitivity in response to a frequency band of a specific region is significantly reduced is improved in a curve shape, thereby maintaining a high overall sensitivity in the human audible frequency band. It is an object to provide a voice recognition sensor.

The present invention detects and detects a sound signal in a form separated by frequency before performing digital sampling and sound signal processing on the spectrum of human speech, so that the speech recognition circuit is higher than a high-power speech recognition sensor based on a conventional microphone, ADC, and DSP circuit. An object of the present invention is to provide a piezoelectric voice recognition sensor that greatly reduces power consumption through simplification.

In addition, the present invention provides a next-generation low-power voice recognition sensor that can replace the microphone sensor portion of a conventional voice sensor composed of a combination of a microphone, ADC, and DSP using a flexible inorganic piezoelectric material with high sensitivity.

In order to solve the above problems, the speech recognition sensor according to an aspect of the present invention includes a plurality of frequency channels having different lengths, and the plurality of frequency channels senses a frequency region overlapping each other, and the The overall form of continuously connecting the outside forms a curved surface, whereby the plurality of frequency channels are evenly frequency-sensed through the frequency range of 200 Hz to 4 kHz, which is the audible frequency range, and at the same time have a high sensitivity response. It is characterized by having characteristics.

The plurality of frequency channels senses voice with higher intensity as the frequency band decreases.

The length ratio between the shortest-length channel sensing the highest frequency region and the longest-length channel sensing the lowest frequency region among the plurality of frequency channels is in a range of 1: 1.5 to 6.5.

The slope of the curved curved surface tends to change more steeply from the shortest channel to the longest channel.

The speech recognition sensor has a resonance shape, and the quality factor of resonance has a value of 35 or less.

The voice recognition sensor according to another aspect of the present invention includes a flexible thin film; A layer of piezoelectric material stacked on the flexible thin film; And an electrode stacked on the piezoelectric material layer, wherein the electrode includes the plurality of frequency separation channels arranged in a line.

The voice recognition sensor further includes a passivation layer stacked in a form that covers the electrode as a whole.

The present invention provides a small voice sensor system to which a voice recognition based Internet of Things (IoT) including the voice recognition sensor is applied.

The present invention provides a smart home appliance including the small voice sensor system.

The present invention provides a wearable electronic device including the voice sensor system.

The voice recognition sensor according to the present invention separates voices sensed through a plurality of frequency-separated channels formed in a curved form through the plurality of channels according to frequency to maintain high sensitivity on the human audible frequency band.

The present invention shows a characteristic of linearly decreasing as the frequency response value through a plurality of frequency separation channels moves from a low frequency region to a high frequency region, thereby exceeding a reference value of a general acoustic diagnosis microphone over the audible frequency region. It shows the response characteristics.

The present invention uses a plurality of frequency-separated channels formed in a curved form to recognize the separated voice signal by converting it from a mechanical vibration signal to an electrical signal through a piezoelectric element, and employs a cochlear sound transmission mechanism in a human body. , By making flexible piezoelectric voice recognition sensor capable of frequency separation and sensor module compatible with it, we realize low-power voice UI for real-time IoT operation.

In addition, by making a flexible piezoelectric material using a multi-channel structure, by separating frequencies by channels and performing voice recognition, using this, the machine can identify languages and speakers in a standby state with the greatest reduction in power consumption. In addition, an embedded voice recognition sensor and module capable of two-way communication and response can be implemented.

The present invention enables faster and more accurate sound signal processing and high-sensitivity recognition by separating and digitally sampling the speech spectrum for each frequency, and the acoustic analysis module can be simplified to reduce costs. This enables speaker identification despite variability such as ambient noise.

In addition, the present invention allows for voice recognition while waiting at all times with little power consumption even in the sleep state.

The present invention makes it easy and convenient to recognize a speaker and a basic command without preliminary operations such as manipulating a voice recognition start and end button.

1 is a comparison diagram showing a difference between a conventional speech recognition system and the present invention.
2 to 10 are cross-sectional views illustrating a method of manufacturing a piezoelectric voice recognition sensor according to an embodiment of the present invention.
11 is a schematic diagram of a piezoelectric voice recognition sensor according to an embodiment of the present invention.
12 and 13, frequency domains set through a plurality of frequency separation channels are displayed on the horizontal axis, and sensitivity is a relative response (dB) on the vertical axis.
14 shows the sensitivity of the conventional microphone and the sensitivity characteristics of the voice sensor according to the present invention.
15 and 16 are views for specifically explaining a voice recognition sensor having a plurality of frequency channels according to the present invention.
17 and 18 are diagrams for specifically explaining a voice recognition sensor having a plurality of frequency channels according to a conventional trapezoidal shape.
19 is a simulation comparison graph with the present invention having a curved surface and a conventional speech sensor having a trapezoid shape.
20 is an image showing a production form of a speech recognition sensor according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in more detail with reference to the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but will be implemented in various different forms, and only these embodiments allow the disclosure of the present invention to be complete, and the scope of the invention to those skilled in the art. It is provided to inform you completely. The same reference numerals in the drawings indicate the same elements.

1 is a comparative view showing the difference between the existing speech recognition system and the present invention. The existing voice recognition system shown at the top of FIG. 1 receives a micro voice signal in analog form, converts it into a digital signal through an analog to digital converter (ADC), and then processes the digital signal through DSP (digital signal processing). Therefore, there is a disadvantage that high power is consumed.

Specifically, the conventional speech recognition technology has a characteristic that is kept constant in a low sensitivity state according to the frequency domain in a state based on a non-resonant type. Conventionally, there is a limitation in that only one output signal is provided through a capacitive method requiring an external DC power source and a dielectric film is used.

On the other hand, the low-power voice recognition sensor having a plurality of frequency channels in the present invention is a piezoelectric sensor having a resonant type, and has a merit that low-power driving is possible because voice recognition is possible immediately. First, the audio signal is separated from a plurality of electrode channels according to frequency, and at the same time, mechanical motion is converted into an electrical signal in a thin film made of a piezoelectric element, whereby an electrical signal is detected in each frequency band.

In other words, in the case of a conventional microphone, high power is consumed because a frequency band filter, ADC, and DSP are used, but the present invention uses a piezoelectric element that separates each frequency and fishes, so power required for a band filter, ADC, or DSP Can be reduced. In addition, since the sensitivity of the microphone is high, the voltage gain to be used in the circuit can be lowered, thereby reducing power consumption and improving circuit stability.

In addition, according to the present invention, a plurality of frequency separation channels are arranged in a line, but in a curved form as a whole, the response characteristics are linearly reduced as the frequency range is shifted from the low frequency region to the high frequency region. In the above process, it is characterized in that the overall sensitivity is maintained from the low frequency 200Hz band to the high frequency 4kHz band.

The present invention uses a piezoelectric method capable of self-driving the sensor unit, and generates a plurality of output signals having different regions in one chip. In addition, it uses unique technology that utilizes flexible inorganic piezoelectric materials.

2 to 10 are cross-sectional views illustrating a method of manufacturing a piezoelectric voice recognition sensor according to an embodiment of the present invention.

Referring to FIG. 2, a silicon substrate 100 as a sacrificial substrate is disclosed. In the present invention, the sacrificial substrate 100 provides stress variation with a metal layer that is later stacked, but is not directly bonded to the nanogenerator device. In one embodiment of the present invention, the compressive stress of the silicon substrate 100 forms a tensile stress and disharmony of the metal layer bonded to the top of the device, and then a separate buffer layer bonded on the silicon substrate 100 by external energy applied. (The silicon oxide layer in one embodiment of the present invention) is cracked, the crack in the horizontal direction of the buffer layer is described in more detail below. In the present invention, in particular, the cracked portion may be adjusted and controlled according to a stress difference between the metal layer and the sacrificial substrate.

A buffer layer 200 such as silicon oxide is stacked on the silicon substrate 100. In the present invention, the buffer layer 200 is at a level that may fall depending on the physical force generated according to the stress difference, and is bonded to the nanogenerator device. In one embodiment of the present invention, a silicon oxide layer is used as the buffer layer 200, and the bonding force between the silicon oxide layer and the nanogenerator can effectively separate the nanogenerator device due to a difference in stress between the lower substrate and the metal layer. Level.

Referring to FIG. 3, the PZT thin film 300 as a piezoelectric material layer is deposited on the buffer layer 200 through a sol-gel process, which is a known technique. In order to remove the organic component from the thin film of sol-gel solution, a 0.4M PZT sol-gel solution (Zr:Ti in a 52:48 molar ratio of PbO greater than 10 mol%) was thermally decomposed in an air atmosphere at 450° C. for 10 minutes. And spin-cast on a wafer at 2500 rpm.

The deposition and pyrolysis steps are repeated several times to form a 2 μm thick PZT thin film. Crystallization of the PZT thin film is carried out in air at 650° C. for 45 minutes. Rapid thermal treatment (RTA) is used for the pyrolysis and crystallization process.

Referring to FIG. 4, a nickel layer 400, which is a metal layer, is stacked on the upper surface of the PZT thin film 300. According to an embodiment of the present invention, the nickel layer 400 may be stacked through a conventional semiconductor process such as sputtering or PVD process, and may also be stacked according to a conventional metal coating method. According to the lamination, nickel 400 bonded to the PZT thin film 300 is formed.

Referring to FIG. 5, mechanical energy (for example, physical impact) or thermal energy is applied to the nickel layer 400 which is the metal layer having the residual tensile stress. As a result, a residual tensile stress of nickel occurs, and a mismatch or asymmetry effect between the residual compressive stress of the silicon substrate and the residual tensile stress of the silicon substrate indirectly bonded to the nanogenerator element through the buffer layer occurs, and thus silicon At the interface between the oxide buffer layer 200 and the PZT thin film 300, the bonding between the two layers falls. The present invention is a metal layer having a tensile stress different from the residual compressive stress of the silicon substrate as described above, and then stacking the desired device and the substrate, and then applying energy from the outside to separate the device from the weak bonding surface. In particular, since the separation surface that generates separation of these devices is set as the interface between the PZT thin film 300 and the buffer layer bonded with the weakest force, the device manufactured on the silicon substrate can be separated and transferred as it is. In addition, the device separation position may be controlled according to a stress difference between the metal layer and the sacrificial substrate.

Referring to FIG. 6, the PZT thin film 300 having a detached bond is separated from the silicon oxide buffer layer 200 according to the residual tensile stress mismatch of the metal layer contacting the silicon substrate (see FIG. 7 ).

Meanwhile, the process of separating the PZT thin film 300 from the silicon oxide buffer layer 200 may be possible by a laser lift off (LLO) process. That is, irradiating the rear surface of the silicon oxide buffer layer 200 through the XeCl-pulse excimer laser to separate the PZT thin film 300 from the buffer layer 200, for example, the photon energy (4.03eV) of the XeCl laser is a buffer layer ( Since it is smaller than the band-gap energy of 200) and larger than that of the PZT thin film 300, it enables the PZT thin film to be transferred to the flexible plastic substrate. As a result, the laser beam penetrates the silicon oxide buffer layer, followed by local melting and dissociation of PZT at the boundary with the buffer layer.

As described above, a laser lift off (LLO) process for converting the PZT thin film into a plastic substrate occurs.

Referring to FIG. 8, the separated PZT thin film 300-nickel 400 layer is physically moved to and bonded to a flexible plastic substrate 600. Thus, the flexible nanogenerator transferred on the flexible plastic substrate 600 is completed.

Referring to FIG. 9, the nickel layer 400 is removed through etching, which is a conventional chemical etching process. For example, the nickel layer 400 may be removed by immersing the upper portion of the device bonded to the plastic substrate 600 in a specific etchant for etching the nickel layer 400. However, in addition to this, the nickel layer 400 may be selectively removed according to various conventional metal layer removal methods, and this is also within the scope of the present invention.

Next, referring to FIG. 10, an electrode 500 is stacked on the PZT thin film 300, thereby forming a flexible plastic film 600 from the bottom, a PZT thin film 300, and an electrode 500. Stacked. Here, the electrode 500 forms a plurality of frequency separation channels.

The electrode 500 may be any one of electrode materials including Ti/Au, Ti/Pt, Cr/Au, and Cr/Pt.

The plastic substrate 600 may be any one of substrate materials including PET, PEN, Parylene, and Kapton.

Meanwhile, referring to FIG. 11, the piezoelectric speech recognition sensor according to the present invention may selectively add a passivation layer in a form that covers the electrode 500 as a whole. The protective layer may be Parylene or SU-8.

An adhesive layer is disposed between the plastic substrate 600 and the PZT thin film 300, and the adhesive layer may be Norland, PU, or the like.

The basic concept of the present invention will be described with reference to FIGS. 12 to 13.

12 and 13, frequency domains set through a plurality of frequency separation channels are displayed on the horizontal axis, and sensitivity is a relative response (dB) on the vertical axis.

As an example, the frequency channels corresponding to the four channels illustrated in FIG. 12 sense frequency regions of regions overlapping each other. That is, on the graph, the leftmost channel 1 shows a high response characteristic, and the rightmost channel 4 shows a relatively low response characteristic.

The present invention can maintain a higher sensitivity in a specific frequency domain by using a resonant element. That is, in contrast to the conventional microphone (Ref. Mic) that maintains a constant sensitivity of about -100dB, it can be seen that all regions overlapping each other in the frequency channel exhibit a state far exceeding the sensitivity of the microphone (Ref. Mic).

Existing microphones consist of a sensor and a read-out IC (ROIC), and when the membrane vibrates due to negative pressure, an electrical signal due to a potential difference enters the ROIC. ROIC consists of an amplifier and an impedance converter, and the sensitivity is determined by the gain of the amplifier.

12 shows the sensitivity of the sensor part excluding the ROIC in ref.mic, and shows response characteristics corresponding to each frequency when white noise of 20-20 kHz is input as 94SPL (sound-level pressure).

For example, in the resonant frequency channel 1 (Ch1), the 3 dB bandwidth of the channel 1 is widened to a level exceeding 50 Hz, and the Q value is set to a direction lowering to less than 35, so that 200 Hz to 4 kHz are evenly sensed.

Next, in FIG. 13, an audible frequency range ranging from 100 Hz to 4000 Hz is set on the horizontal axis, and sensitivity is a relative response (dB) on the vertical axis.

The frequency domain set through the multiple frequency separation channels is displayed and the vertical axis shows the relative response (dB) sensitivity.

In FIG. 13, sensitivity obtained through 7 frequency separation channels is shown.

Compared to the conventional microphone (Ref. Mic), which maintains the sensitivity constantly through the entire frequency band, the response performance through multiple frequency separation channels within the frequency range from 125 Hz to 4000 Hz is the sensitivity of the microphone (Ref. Mic). You can see that it shows a state of overshoot.

The plurality of frequency channels senses a frequency region of a region overlapping each other, and the plurality of frequency channels senses voice with higher intensity as the frequency band decreases.

On the other hand, referring to Figure 14, the sensitivity of the existing microphone is to indicate the output signal from 94dB (Sound Pressure Level, SPL) conditions at 1kHz, the voice sensor according to the present invention is about the same as the measurement conditions of the microphone It had a peak-to-peak (30 mV) value, which represents a value of about 10.6 mV (rms). When the 10.6mV (rms) is converted, it has -39.5dBV, through which the sensitivity of the device according to the present invention is -39.5dBV. Meanwhile, a high sensitivity sensor having a margin of -45 dBV or more is provided.

The above sensitivity conversion calculation is possible through the Internet address "http://www.sengpielaudio.com/calculator-db-volt.htm".

As described above, the present invention enables realization of a plurality of sensitive areas using a multi-channel, and achieves high sensitivity in a range of 200 Hz to 4 kHz corresponding to a human voice band.

Hereinafter, a voice recognition sensor having a plurality of frequency channels according to the present invention will be described in detail with reference to FIGS. 15 and 16.

The speech recognition sensor has a plurality of frequency channels having different lengths. The plurality of frequency channels may be in the form of six electrode channels arranged to have a predetermined interval on one chip. The plurality of frequency channels are arranged side by side in the order of ch6 by ch1 from left to right on the upper picture in FIG. 14.

When the plurality of frequency channels are continuously connected to both side edges, the overall shape forms a curved surface.

Specifically, when going from ch1 to ch2 and ch3, the slope increases to a fairly gentle state, and when going from ch3 to ch4, ch5, ch6, it can be seen that it increases to a state having a significantly large slope value.

That is, the slope value of the curved surface tends to gradually increase from the shortest-length channel ch1 to the longest-length channel ch6.

On the lower graph of FIG. 15, it can be seen that the frequency sensing region is uniform as the channel length from the shortest channel ch1 to the longest channel ch6 increases.

15 and 16 show the tendency based on the highest peak point in each frequency channel. Here, the vibration signal and the electrical signal coincide in almost all frequency channels.

Among the six electrode channels, a frequency channel on one side for sensing a low frequency is set as a first sensing unit, and a frequency channel on the other side for sensing a high frequency relative to the first sensing unit is set as a second sensing unit.

The length ratio between the second sensing unit corresponding to the shortest length channel sensing the highest frequency region and the first sensing unit corresponding to the longest length channel sensing the lowest frequency region is in the range of 1:1.5 to 6.5. .

Preferably, the ratio between the shortest length channel and the longest length channel may be in the range of 1:4. This is due to the fact that in the case of 1:3, the frequency cover range may be biased toward the high frequency, and in the case of 1:5, bias may be biased toward the low frequency.

When the ratio of the length of the electrode channel decreases on the first sensing unit is set to a, and when the ratio of the length of the electrode channel decreases on the second sensing unit is set to b,

The ratio between a and b, b/a is preferably determined in a range smaller than 1.

Through this, the performance of sensing the low frequency in the first sensing unit in which the ratio in which the length of the electrode channel decreases is set to a may be further improved.

Hereinafter, a voice recognition sensor having a plurality of frequency channels according to a conventional trapezoidal shape will be described with reference to FIGS. 17 and 18.

When a plurality of frequency channels having different lengths are continuously connected to both side edges, the overall shape is trapezoidal.

Specifically, when going from ch1 to ch6, it can be seen that the slope is constant.

That is, the channel length tends to increase at a constant rate from the shortest-length channel ch1 to the longest-length channel ch6.

Looking at the bottom graph of FIG. 17, it is shown that there is no clear boundary in the frequency domain to be detected when the longest channel of ch6 fluctuates from ch2. That is, while the maximum peak point of the frequency decreases finely from ch6 to ch4, it is seen that it increases significantly from ch4 to ch3 and then increases from ch3 to ch2 finely. On the other hand, it can be seen that a frequency region that is suddenly detected in the shortest-length channel ch1 detects a remarkably high state, and a frequency region that cannot be covered between ch2 and ch1 has occurred.

As described above, a conventional sensor having a trapezoidal shape has a problem in that a frequency region not covered by a plurality of frequency channels is remarkably exposed.

Referring to FIG. 19, a simulation comparison graph with the present invention having a curved surface and a conventional voice sensor having a trapezoid shape is shown.

The horizontal axis shows a state in which a plurality of frequency channels are arranged according to a distance, and the vertical axis shows a resonant frequency according to a plurality of separated frequency channels.

When the conventional voice sensor follows the resonance frequency equation in the trapezoidal shape, it can be confirmed that it has a parabolic resonance frequency arrangement.

On the other hand, it can be seen that the present invention separates the resonant frequency in a state almost close to a straight line through a curved surface.

The voice sensor according to the present invention is a piezoelectric type resonant element, and the quality factor corresponds to an important feature.

Q (Quality factor) at resonance refers to the quality of frequency selective characteristics. The difference between the frequencies of the points at which the resonance frequency is 3 dB on both sides, that is, attenuated in half, is called the 3 dB bandwidth, and the Q value is the resonance frequency divided by the 3 dB bandwidth. In other words, the sharper the resonance characteristic, the narrower the 3dB bandwidth will be, and eventually the Q value will be larger. On the other hand, if Q is low, it may mean that the band is wide.

In the present invention, the 3dB bandwidth of each channel is set to a direction in which the Q value is lowered by widening the bandwidth to 50Hz or more, so that it is evenly sensed with a higher sensitivity than the conventional MEMS-based microphone within the range of 200Hz to 4kHz. Existing MEMS-based microphone is the performance of a pure microphone sensor excluding the effect of the ROIC part, which means -100dB of performance at a white noise condition of 20 to 20,000 Hz of about 94SPL, and the flexible piezoelectric-based voice recognition sensor according to the present invention is 200 to It means that it has a performance higher than -100dB, which represents the performance value of a conventional MEMS-based microphone in the range of 90% or more of 4kHz.

The electromechanical coupling factor K and the electromechanical quality factor Q show more effective performance than the piezoelectric coefficient.

The quality factor includes electrical quality factor (Qe) and mechanical quality factor (Qm).

Qe represents the reciprocal of the electrical loss (tanδ), while Qm represents the concentration of displacement against stress due to mechanical vibration damping of the vibrating body. The piezoelectric material for resonators requires a Qm value of 1500 or higher, while the filter uses a Qm value of 400 to 600 or so, and a piezoelectric speaker focuses on material development with a Qm value of 80 or lower and high dielectric constant. have.

In the case of the voice sensor according to the present invention, a multi-channel may have a total of seven channels, and the Q value has a value between 18 and 28. Based on the above results, it may be desirable to keep the Q value below 35.

The mechanical quality factor indicates the ratio of energy accumulated during exchange between electrical energy and mechanical energy, and is due to a phase difference from an applied voltage generated when the permanent dipoles move. Loss is mostly dissipated in the form of thermal energy, It determines the sharpness of resonance caused by the piezoelectric body at the resonance frequency.

When the value of the mechanical quality factor is low, degradation generally occurs quickly and has a different value from the electrical quality factor.

20 is an image showing a production form of a speech recognition sensor according to an embodiment of the present invention.

Referring to FIG. 20, PZT, a piezoelectric material, is laminated on a PCB having an empty space using PET. The piezoelectric material forms a plurality of frequency channels formed in a curved shape and is formed to correspond to the empty space. Meanwhile, the Au electrode is formed on the upper surface of the PCB through a process of being electrically connected to both ends of the piezoelectric material.

PZT and PU adhesives are placed on PET, which is a square-shaped transparent plastic substrate, and electrical energy generated from PZT is collected through Au electrodes. In addition, a protective layer (Passivation layer) that protects it is additionally deposited to protect the device.

On the other hand, PET, UV-sensitive PU adhesive, PZT, protective layer may be made of a transparent material. In the Au electrode, the color of the electrode may be visible to the naked eye.

The present invention focuses on realizing speech recognition by simulating the cochlear, which is a human auditory organ, and it is possible to significantly reduce power consumption with a simple circuit based on a flexible piezoelectric voice sensor rather than a conventional microphone, ADC, and DSP combination method for frequency separation. have. In addition, if an efficient recognition algorithm compatible with this is implemented, the natural language of a human can be recognized with high selectivity, sensitivity, detection speed, and stability.

This technology can be applied to real life. For example, it is possible to safely use the vehicle information system by voice while driving, only by voice while in standby mode. Only low-power control can be achieved. In particular, it is possible to use facilities such as elevators more conveniently by registering caring or voices for the handicapped and patients with disabilities and limbs.

This technology is a theme that encompasses all IT-NT-BT-materials, and is a convergent technology that enriches human life by taking inspiration from nature. Through the voice of the speaker, it is possible to grasp the identity, psychology, health status, language ability, etc. in the standby mode with low power, enabling personalized service provision, and covering all areas of the sensor, including security, finance, and medical education. It can be utilized in.

In particular, it is possible to apply to mobile healthcare, such as analyzing the emotional state by detecting and analyzing and storing voice patterns in big data and deriving psychological stability through a feedback system, and a security system through voice authentication and speaker identification. It is expected to be strengthened to help protect personal information and privacy.

The present invention can implement an ultra-small voice sensor system for Internet of Things (IoT) and mobile based on voice recognition through the above features.

The voice sensor according to the present invention can be used in smart home appliances including TVs and refrigerators, voice assistants, and voice security applications.

In the present invention, the mechanical recognition energy generated by the voice is separated into different positions for each frequency by using the piezoelectric material before the digital sampling and sound signal processing of the human voice spectrum by the speech recognition sensor made of a high-efficiency inorganic piezoelectric material on a flexible substrate. Afterwards, it is converted into an electrical signal and the voice signal is processed in parallel for each frequency.

In the present invention, a plurality of frequency separation channels form a shape of an artificial cochlea that resembles a xylophone shape, and as the sizes of the plurality of frequency separation channels are changed, locations where high and low frequency resonances resonate change to physically reproduce human voice. To separate. Here, each separated sound is amplified through an analog circuit for each frequency, filtered and converted into a digital signal for processing. In this process, power consumption is significantly reduced compared to the conventional microphone, ADC, and DSP combination.

The present invention provides a piezoelectric voice recognition sensor coupled on a flexible thin film, and can be used even when attached to clothing. That is, it is possible to apply the technology to harvest the physical energy of the sound wave and ultrasonic regions easily generated in the surroundings while attached to the garment and convert it into electrical energy.

In general, in order to realize a ubiquitous network ``everywhere'', the existence of a ubiquitous power source ``everywhere and operating'' is indispensable. On the other hand, the power of ubiquitous network components that exist everywhere should be of a self-sufficient form that does not require charging. In other words, power generation and power storage must be provided.

As described above, the piezoelectric voice recognition sensor according to the present invention separates the voice detected using the plurality of frequency separation channels formed in a trapezoidal form through the plurality of channels at the same time and piezoelectrically separates the separated voice signals. It converts the mechanical vibration signal to an electrical signal through the device for recognition.

Although preferred embodiments of the present invention have been described above, the present invention is not limited to the specific embodiments described above. That is, a person having ordinary knowledge in the technical field to which the present invention pertains can make a number of changes and modifications to the present invention without departing from the spirit and scope of the appended claims, and all such appropriate changes and modifications Equivalents should also be considered within the scope of the present invention.

Claims (13)

Flexible piezoelectric based speech recognition sensor,
The speech recognition sensor has a plurality of frequency channels that are spaced apart from each other with a different length,
The plurality of frequency channels sense a frequency region overlapping each other,
A voice recognition sensor characterized by having a response characteristic having a high sensitivity of -45dBV or higher at 94dB(SPL) of 1kHz through the entire shape of a curved surface in which the upper and lower ends of the plurality of frequency channels are continuously connected to each other.
According to claim 1,
The plurality of frequency channels are evenly sensed through the frequency range of 200 Hz to 4 kHz, which is the audible frequency range, and at the same time, the voice is sensed with higher intensity as it goes to a lower frequency band.
Speech recognition sensor.
delete According to claim 1,
The length ratio between the shortest-length channel sensing the highest frequency region and the longest-length channel sensing the lowest frequency region among the plurality of frequency channels is in the range of 1: 1.5 to 6.5,
Speech recognition sensor.
The method of claim 4,
The slope of the curved surface gradually changes from the shortest channel to the longest channel,
Speech recognition sensor.
Flexible piezoelectric based speech recognition sensor,
The speech recognition sensor has a plurality of frequency channels having different lengths,
The plurality of frequency channels sense a frequency region overlapping each other,
The speech recognition sensor has a resonance type, and the quality factor of each channel has a value of 35 or less,
The quality factor widens the 3dB bandwidth of each channel to 50Hz or more,
Using more than 90% of the 200-4kHz frequency range, using a piezoelectric material with high sensitivity characteristics, more sensitive sensing and even sensing when compared with conventional MEMS-based microphones that sense sound pressure with a change in the capacitance of the microphone sensor Characterized by a speech recognition sensor.
delete The method of claim 6,
The conventional commercially available MEMS-based microphone has the sensitivity of the microphone sensor itself, except for the voltage gain of the ROIC, of about 94SPL and -100dB in a white noise environment of 20-20kHz, whereas the flexible piezoelectric sensor has a 90 of 200~4kHz. Characterized in that it has a sensitivity of -100dB or more within the% range,
Speech recognition sensor.
According to claim 1,
Flexible thin film;
A layer of piezoelectric material stacked on the flexible thin film; And
It includes; an electrode stacked on the piezoelectric material layer,
The electrode is in a state including the plurality of frequency separation channels arranged in a line,
Speech recognition sensor.
The method of claim 9,
The voice recognition sensor,
Further comprising a protective layer (Passivation layer) laminated in a form that covers the electrode as a whole,
Speech recognition sensor.
Claim 1 to claim 2, claim 4 to claim 6 and comprising the speech recognition sensor according to any one of claims 8 to 10,
Small voice sensor system with voice recognition based Internet of Things (IoT).
Including the voice sensor system according to claim 11,
Smart home appliances.
Including the voice sensor system according to claim 11,
Wearable electronic device.

Images