A frequency adaptive triboelectric artificial larynx sensor and human-computer interaction speech recognition system

Abstract

The application discloses a frequency adaptive triboelectric artificial larynx sensor and a man-machine interactive voice recognition system, and relates to a laryngeal voice collection sensor. The sensor comprises a friction layer, a gradient composite functional layer, an interdigital electrode layer and a flexible substrate. The gradient composite functional layer is a single-layer polymer film, the inside of which is gradiently doped with nanofillers with different functions along the thickness direction, and is embedded with a biomimetic microcavity array. The vertical gradient design makes the film sensitive to vibration of a specific frequency band at different depths, and the biomimetic microcavity amplifies the target frequency signal through mechanical resonance. The two cooperate to realize frequency selection and signal enhancement of the hardware end in the single-layer structure, effectively solving the problems of interlayer interference, large thickness and low signal-to-noise ratio existing in the traditional multi-layer structure. The sensor has the advantages of lightness, high fidelity and high signal-to-noise ratio, and is suitable for voice recognition and synthesis, laryngeal health monitoring and new man-machine interaction.

Description

Technical Field

[0001] This invention relates to the fields of flexible electronics, biomedical sensing, and human-computer interaction, specifically to a frequency-adaptive triboelectric artificial larynx sensor and a human-computer interaction speech recognition system. Background Technology

[0002] The restoration of speech function in aphoniac patients (such as those who have undergone laryngectomy) is a significant social need. Non-invasive artificial laryngeal sensors based on laryngeal surface vibration detection are one of the key technologies. Triboelectric nanogenerators show promise in this field due to their high sensitivity, self-powered characteristics, and wide range of material choices.

[0003] In existing technologies, to improve the response to different frequency speech components (such as low-frequency vowels and high-frequency consonants), a multi-layered thin-film stacked structure with different inherent frequencies is often used. However, this structure has inherent drawbacks: First, multi-layer stacking increases the overall thickness of the sensor, affecting wearing comfort and concealment; second, severe mechanical coupling and signal crosstalk between layers make it difficult to clearly separate characteristic signals of different frequency bands, resulting in a low signal-to-noise ratio; finally, multi-layer alignment and packaging processes are complex, making it difficult to guarantee manufacturing consistency.

[0004] Therefore, there is an urgent need to develop a new type of artificial laryngeal sensor that is structurally integrated, thin and light, and can achieve frequency-selective response and high signal-to-noise ratio signal output at the hardware level. Summary of the Invention

[0005] This invention aims to overcome some shortcomings of existing technologies and provide a frequency-adaptive triboelectric artificial laryngeal sensor based on a vertically gradient composite thin film. Its core lies in integrating frequency selection functionality within a single-layer thin film. Through a vertically gradient material design combined with a built-in microcavity resonant structure, it achieves high-fidelity, high signal-to-noise ratio sensing of complex laryngeal vibration modes.

[0006] To achieve the above objectives, in a first aspect of the present invention, a frequency-adaptive triboelectric artificial laryngeal sensor is provided, comprising a triboelectric layer, a gradient composite functional layer, an interdigital electrode layer and a flexible substrate disposed from top to bottom;

[0007] The gradient composite functional layer is a single-layer polymer-based composite film with functional fillers distributed vertically along the thickness direction inside, and a non-penetrating biomimetic microcavity array embedded inside the film; a friction gap is provided between the friction layer and the gradient composite functional layer.

[0008] The vertical gradient distribution is as follows: the upper surface region near the friction layer is doped with one-dimensional conductive nanomaterials with high aspect ratio and two-dimensional insulating and thermally conductive nanosheets with high aspect ratio; the main body region in the middle of the film is uniformly dispersed with nanoparticles with high dielectric constant; and the lower bottom layer region near the interdigitated electrode layer is doped with low-density microspheres or three-dimensional conductive network materials.

[0009] The size of the biomimetic microcavity array is designed so that its inherent resonant frequency matches the human speech audio frequency range.

[0010] The interdigitated electrode layer is fabricated on the lower surface of the gradient composite functional layer.

[0011] In one specific embodiment, the shape of the microcavities in the biomimetic microcavity array is at least one of cylindrical, conical, or hemispherical, and the depth of the microcavities varies in a gradient from the central region of the film to the edge region, in order to correspond to the spatial distribution differences of different vibration modes on the throat surface.

[0012] In one specific embodiment, in the vertical gradient distribution, the mass fraction of the one-dimensional conductive nanomaterial in the upper surface region is 0.05%-0.5%, the mass fraction of the two-dimensional insulating and thermally conductive nanosheet is 0.01%-0.2%, the mass fraction of the high dielectric constant nanoparticles in the middle main region is 5%-20%, and the mass fraction of the low-density microspheres or three-dimensional conductive network material in the lower bottom layer region is 2%-10%.

[0013] In one specific embodiment, the pattern of the interdigital electrode layer is a plurality of concentric ring interdigital electrodes that are electrically insulated from each other, and the projection position of each ring interdigital electrode on the plane corresponds to the microcavity regions of different sizes or different distribution densities in the biomimetic microcavity array.

[0014] In one specific embodiment, it further includes an integrated flexible encapsulation layer, which completely encapsulates the friction layer, the gradient composite functional layer, the interdigitated electrode layer, and the flexible substrate. The encapsulation portion above the friction layer is a stretchable breathable film, and the encapsulation portion below the flexible substrate is a biocompatible pressure-sensitive adhesive layer.

[0015] In one specific embodiment, the friction layer is a polymer film with a micro-nano composite structure on its surface. The micro-nano composite structure includes micron-sized protrusions formed by a template method and nanowires or nanoparticles grown in situ on the surface of the protrusions.

[0016] In one specific embodiment, the vertical gradient distribution is formed in one step by a gradient centrifugal casting method, wherein different fillers spontaneously form a continuous gradient distribution along the thickness direction under the centrifugal force field due to density differences.

[0017] In one specific embodiment, the gradient composite functional layer is a polydimethylsiloxane with a gradient degree of crosslinking, wherein the degree of crosslinking of the upper surface region is higher than that of the lower bottom layer region, so as to provide surface rigidity to facilitate high-frequency vibration transmission and bottom layer flexibility to facilitate the capture of low-frequency large-amplitude vibrations.

[0018] In one specific embodiment, the inner wall of the biomimetic microcavity is modified with a nano-coating with a dielectric constant different from that of the substrate material, which is used to regulate the local electric field distribution and equivalent dielectric constant of the microcavity to optimize its triboelectric output performance.

[0019] In one specific embodiment, the sensor further includes a signal preprocessing circuit module integrated on the flexible substrate. The signal preprocessing circuit module is electrically connected to the interdigitated electrode layer and is used to perform differential amplification, filtering, and impedance transformation on the multi-channel signals.

[0020] In a second aspect of the present invention, a human-computer interaction voice recognition system is provided, comprising:

[0021] The artificial laryngeal sensor provided in the first aspect of the present invention;

[0022] The signal acquisition module is used to acquire the pulse current signal of the interdigital electrode layer;

[0023] The signal processing module is used to calculate the throat vibration identification signal based on the pulse current signal.

[0024] Beneficial effects: 1) Structural integration and thinning: By integrating a vertical gradient design within a single-layer thin film with microcavities, the traditional multi-layer stacked structure is replaced, significantly reducing sensor thickness and improving wearing comfort. 2) Hardware-side frequency selectivity and high signal-to-noise ratio: The vertical gradient structure makes different depth regions of the thin film sensitive to specific frequency band vibrations (high frequency at the surface, low frequency at the bottom), while the biomimetic microcavities amplify specific frequency signals through mechanical resonance. The synergy of these two elements is equivalent to integrating a "mechanical filter" within the material, improving the signal-to-noise ratio of the target frequency band signal from the source and more clearly characterizing complex vibration modes such as bubble sounds and aerodynamic sounds. 3) Rich signal dimensions: Combining gradient functional layers with patterned interdigitated electrodes, differentiated acquisition of charge signals in spatial (planar position) and depth (vertical direction) dimensions can be achieved, providing richer feature information for backend signal processing. 4) High manufacturing feasibility: It can be prepared using controllable processes such as gradient centrifugal casting and layer-by-layer spin coating, is compatible with microfabrication technology, and is conducive to large-scale production. 5) Strong functional expandability: This sensor can not only be used for speech recognition and synthesis, but its high-fidelity vibration sensing capability also makes it widely applicable in the fields of laryngeal health monitoring (such as vocal cord tremor analysis and swallowing function assessment) and new human-computer interaction interfaces. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the overall structure of a sensor according to an embodiment of the present invention;

[0026] Figure 2 This is a schematic diagram of the gradient composite functional layer of a sensor according to another embodiment of the present invention. Detailed Implementation

[0027] The embodiments of this patent are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this patent, and should not be construed as limiting this patent.

[0028] Example 1

[0029] like Figures 1-2 As shown, the first embodiment of the present invention provides a frequency-adaptive triboelectric artificial laryngeal sensor, including a triboelectric layer 100, a gradient composite functional layer 200, an interdigital electrode layer 300 and a flexible substrate 400 arranged from top to bottom;

[0030] The gradient composite functional layer 200 is a single-layer polymer-based composite film 201, which has functional fillers with a vertical gradient distribution along the thickness direction, and a non-penetrating biomimetic microcavity array 202 is embedded inside the film; a friction gap 500 is provided between the friction layer 100 and the gradient composite functional layer 200.

[0031] The vertical gradient distribution is as follows: the upper surface region near the friction layer 100 is doped with one-dimensional conductive nanomaterials with high aspect ratio and two-dimensional insulating and thermally conductive nanosheets with high aspect ratio; the main body region in the middle of the film is uniformly dispersed with nanoparticles with high dielectric constant; the lower bottom layer region near the interdigitated electrode layer 300 is doped with low-density microspheres or three-dimensional conductive network materials.

[0032] Typically, the upper surface region is a high-density AgNWs / BNNS region 203, containing one-dimensional conductive nanomaterials AgNWs and two-dimensional insulating and thermally conductive nanosheets BNNS. Specific materials can be selected according to actual needs, and this invention does not impose specific limitations. High-dielectric-constant nanoparticles can be BST, distributed in a uniformly distributed BST region 204; low-density microspheres or three-dimensional conductive network materials can be hollow glass microspheres 205. In practical applications, it is not necessary to strictly prevent the materials in each of the three layers from interfering with other layers; it is only necessary to ensure the dominant mixing in that region. For example, the main mixing material in the upper surface region is one-dimensional conductive nanomaterials and two-dimensional insulating and thermally conductive nanosheets, but a small amount of high-dielectric-constant nanoparticles and low-density microspheres can be mixed in.

[0033] The size of the biomimetic microcavity array 202 is designed so that its inherent resonant frequency matches the human speech audio segment.

[0034] The interdigitated electrode layer 300 is fabricated on the lower surface of the gradient composite functional layer 200.

[0035] In this embodiment, by integrating "mechanical filtering" (microcavity resonance) and "electrical filtering" (gradient sensitive region) within a single medium, the throat vibration of the target frequency band can be preferentially responded to and amplified at the physical level, directly outputting a raw signal with a higher signal-to-noise ratio, thus reducing the burden and complexity of back-end signal processing. Compared to traditional layered structures, the single-layer thin-film design reduces the thickness of the sensor's active layer to the hundreds of micrometers, greatly improving the invisibility, comfort, and consistency with skin deformation, and avoiding signal attenuation and motion artifacts caused by a thick structure.

[0036] Optionally, in the biomimetic microcavity array 202, the shape of the microcavity is at least one of cylindrical, conical, or hemispherical, and the depth of the microcavity varies in a gradient from the central region of the film to the edge region, in order to correspond to the spatial distribution differences of different vibration modes on the throat surface.

[0037] In this embodiment, by matching the depth of the microcavity to the typical vibration frequencies of different functional areas of the larynx (such as the vocal cord projection area and the cartilage vibration area), "spatial programming" of the sensor's sensing performance is achieved, enabling more accurate capture of characteristic vibration patterns of different articulation methods (such as true voice, breathy voice, and bubbly voice). The depth-gradient microcavity array allows the same vibration event to elicit differentiated resonant responses at different locations on the sensor, providing multi-dimensional and complementary mechanical feature signals for a single speech unit, greatly enhancing the feature library for pattern recognition.

[0038] In this embodiment, in the vertical gradient distribution, the mass fraction of the one-dimensional conductive nanomaterial in the upper surface region is 0.05%-0.5%, the mass fraction of the two-dimensional insulating and thermally conductive nanosheet is 0.01%-0.2%, the mass fraction of the high dielectric constant nanoparticles in the middle main region is 5%-20%, and the mass fraction of the low-density microspheres or three-dimensional conductive network material in the lower bottom layer region is 2%-10%.

[0039] In this embodiment, the concentration range is a "golden range" verified through extensive experiments. Below the lower limit, the functional effect is not obvious; above the upper limit, the filler is prone to agglomeration, leading to deterioration of the film's mechanical properties, conductive permeation, or a surge in dielectric loss. This limitation ensures the optimal balance between sensor sensitivity, flexibility, and reliability. It provides quantifiable and repeatable formulation standards for manufacturing, ensuring consistency and yield between product batches, and is a key support for the technology to move from the laboratory to industrialization.

[0040] In this embodiment, the pattern of the interdigital electrode layer 300 is a plurality of concentric ring interdigital electrodes that are electrically insulated from each other, and the projection position of each ring interdigital electrode on the plane corresponds to the microcavity regions of different sizes or different distribution densities in the biomimetic microcavity array 202.

[0041] Different ring electrodes collect charges from microcavity regions with different resonant characteristics, enabling the output signal to carry a spatial-frequency label at the hardware level. This is equivalent to performing preliminary signal pre-classification within the sensor, providing a cleaner and easier-to-decode multi-channel input for backend algorithms. Common-mode environmental noise (such as body motion) is collected approximately synchronously by each ring electrode, while specific vibration signals originating from a specific location in the throat are highlighted only on that particular electrode. Using techniques such as differential amplification, common-mode noise can be suppressed more effectively.

[0042] The artificial laryngeal sensor in this embodiment also includes an integrated flexible encapsulation layer, which completely encapsulates the friction layer 100, the gradient composite functional layer 200, the interdigitated electrode layer 300, and the flexible substrate 400. The encapsulation portion above the friction layer 100 is a stretchable breathable film, and the encapsulation portion below the flexible substrate 400 is a biocompatible pressure-sensitive adhesive layer.

[0043] In this embodiment, the upper breathable film protects the friction layer 100 while allowing sweat evaporation, preventing stuffiness and skin discomfort; the lower biocompatible pressure-sensitive adhesive ensures a tight and stable fit between the sensor and the skin, preventing it from easily falling off or slipping even during high-dynamic activities, thus guaranteeing long-term continuity and stability of signal acquisition. The excellent wearing comfort and discreetness encourage users to wear it for extended periods, which is crucial for continuous health monitoring (such as sleep apnea analysis and vocal fatigue assessment).

[0044] Optionally, the friction layer 100 is a polymer film with a micro-nano composite structure on its surface, the micro-nano composite structure including micron-scale protrusions formed by a template method and nanowires or nanoparticles grown in situ on the surface of the protrusions.

[0045] In this embodiment, the combination of micron-level protrusions and nanoscale structures creates a fractal rough surface, resulting in an order-of-magnitude increase in contact area. Under the same vibration, the contact separation process is more complete, and the amount of charge transfer is significantly improved, especially effectively reducing the detection limit for weak high-frequency vibrations (such as consonants and aerophones). Material synergy effect: The in-situ grown nanowires / particles (such as ZnO) may themselves possess piezoelectric or semiconductor properties, which can synergize with the triboelectric effect of the substrate, further enhancing output performance.

[0046] Optionally, the vertical gradient distribution is formed in one step by a gradient centrifugal casting method, wherein different fillers spontaneously form a continuous gradient distribution along the thickness direction under the centrifugal force field due to density differences.

[0047] In this embodiment, the filler density difference spontaneously forms a continuous transition distribution in a centrifugal force field, avoiding the obvious interfaces that may occur with layer-by-layer coating. This ensures a smooth transition of the mechanical and electrical properties within the film, resulting in a more continuous and abrupt frequency response. The one-step molding of the core functional layer reduces production steps and lowers the cumulative errors and costs caused by multiple alignments and curing processes, making it particularly suitable for the potential needs of large-scale roll-to-roll production.

[0048] Optionally, the gradient composite functional layer is a polydimethylsiloxane with a gradient change in crosslinking degree, wherein the crosslinking degree of the upper surface region is higher than that of the lower bottom layer region, so as to provide surface rigidity to facilitate high-frequency vibration transmission and bottom layer flexibility to facilitate the capture of low-frequency large-amplitude vibrations.

[0049] The higher degree of cross-linking (modulus) on the surface layer facilitates the efficient transfer of energy from high-frequency micro-vibrations to the internal functional filler; the lower degree of cross-linking (modulus) on the bottom layer makes it more susceptible to large deformation under low-frequency, large-amplitude vibrations, thus more effectively driving microcavity resonance and charge separation. This gradient matching of mechanical impedance maximizes the energy capture efficiency across the entire frequency band. The high degree of cross-linking on the surface layer also enhances the wear resistance and tear resistance of the film surface, extending the sensor's lifespan.

[0050] In this embodiment, the inner wall of the biomimetic microcavity is modified with a nano-coating with a dielectric constant different from that of the substrate material, which is used to regulate the local electric field distribution and equivalent dielectric constant of the microcavity to optimize its triboelectric output performance.

[0051] In this embodiment, the local dielectric environment of the microcavity is altered by a modification layer, optimizing the electron cloud distribution and binding ability of the triboelectric material pairs. When the microcavity resonates, the charge generation, transfer, and collection processes at the inner wall interface are significantly enhanced, allowing the energy of mechanical resonance to be converted into a measurable electrical signal more efficiently, amplifying the electrical performance of the "resonance amplification" effect. By selecting modification layer materials with different dielectric constants (such as alumina, titanium dioxide, etc.), the "electrical resonance" characteristics of the microcavity can be flexibly controlled, providing a new dimension for fine-tuning sensor performance.

[0052] Optionally, the sensor further includes a signal preprocessing circuit module integrated on the flexible substrate 400. The signal preprocessing circuit module is electrically connected to the interdigitated electrode layer 300 and is used to perform differential amplification, filtering, and impedance transformation on multi-channel signals.

[0053] Working principle

[0054] The core power generation mechanism of the sensor is the working principle of a triboelectric nanogenerator, which is based on the combination of two physical phenomena: 1) Contact electrification: When two different materials come into contact and separate, due to their different affinities for electrons (work functions), electrons will transfer from one material to the other, making one material negatively charged and the other positively charged. 2) Electrostatic induction: When a charged object approaches a conductor, the free charges inside the conductor will redistribute under the influence of an electric field, inducing opposite charges at the end closer to the charged object and like charges at the end farther away. In this sensor, the triboelectric layer (such as nylon) and the gradient composite functional layer (PDMS / BST matrix) constitute a triboelectric material pair. Typically, nylon easily loses electrons and becomes positively charged, while PDMS easily gains electrons and becomes negatively charged (according to the triboelectric series).

[0055] Assuming an initial state where the friction layer and functional layer are in close contact (the actual initial state could also be a separated state): 1) Contact phase: Vibration of the throat muscles and skin pushes the friction layer and functional layer to compress and fully contact each other. At the contact interface, due to the contact electrification effect, electrons transfer from nylon (friction layer) to PDMS (functional layer), making the upper surface of the functional layer negatively charged and the lower surface of the friction layer positively charged. 2) Separation phase: Vibration reverses, and the two begin to separate. Since the charges are bound to the insulating surface (PDMS and nylon are both insulators), the positive and negative charges cannot be neutralized immediately, thus forming a vertical separation gap and potential difference between the two layers. 3) Electrostatic induction and current generation: This potential difference drives the flow of free electrons in the bottom interdigitated electrodes. To balance this electric field, electrons flow from one electrode to the other (flowing between the interdigitated electrodes), thus generating a momentary pulse current in the external circuit. 4) Reverse contact phase: When vibration brings the two into contact again, the potential difference decreases, electrons flow back, generating a reverse current pulse. 5) This cycle repeats, and the continuous mechanical vibration of the throat is converted into an alternating current signal output.

[0056] This invention utilizes a vertical gradient and biomimetic microcavity design to directionally enhance and frequency-select this fundamental physical process. The vertical gradient structure allows materials at different depths to "perform their respective functions."

[0057] Surface layer (AgNWs / BNNS region):

[0058] Function: Responsible for capturing high-frequency, weak vibrations (such as breath sounds and consonants).

[0059] Principle: The highly conductive AgNWs network forms local "microelectrodes," enabling more efficient collection and transfer of charges generated at the contact interface. BNNS improves thermal conductivity and surface "hardness," making the surface layer sensitive to minute and rapid deformations, resulting in more efficient electron transfer.

[0060] Bottom layer (HGMs / MWCNTs area):

[0061] Function: Specializes in low-frequency, large-amplitude vibrations (such as vowels and vocal fry).

[0062] Principle: HGMs act like tiny springs, reducing the stiffness of the local material and making the area more susceptible to large deformation under low-frequency, high-amplitude vibrations. Greater deformation means greater contact / separation distances and more intense charge separation, resulting in a stronger electrical signal. The MWCNTs network acts like a high-efficiency highway, ensuring that the charges generated in these underlying layers are quickly collected to the bottom electrode, reducing losses.

[0063] Intermediate layer (BST area):

[0064] Function: As the "main force", it efficiently converts medium-frequency energy.

[0065] Principle: BST nanoparticles with high dielectric constant can significantly enhance the capacitance effect of composite films. According to Q=C... V (charge = capacitance) Under the same induced voltage V, a larger capacitor C can bind and output more charge Q, thereby significantly amplifying the output signal strength.

[0066] This invention utilizes a biomimetic microcavity array to achieve built-in "mechanical filters" and "amplifiers," based on the core principle of mechanical resonance. Each microcavity acts like a tiny eardrum, possessing its own inherent resonant frequency. When the frequency of throat vibration coincides with or approaches the resonant frequency of a microcavity, it causes intense vibration of the microcavity wall. This enhances the triboelectric effect in the following ways: 1) Local deformation amplification: The amplitude of the resonant vibration of the microcavity wall is much greater than the uniform deformation amplitude of the overall film. This leads to a more intense contact-separation process between the friction layer and the functional layer in this region. 2) Enhanced charge separation: More intense separation means a larger instantaneous charge separation distance and a stronger local electric field, resulting in a significantly amplified peak electrical signal at that specific frequency. 3) Frequency selectivity: The microcavity responds weakly to vibrations at non-resonant frequencies, thus these frequency components are not amplified. This is equivalent to integrating countless mechanical bandpass filters within the material, directly highlighting key frequency components (such as the fundamental frequency and resonant peaks) in the speech signal at the physical level, greatly improving the signal-to-noise ratio and feature clarity of the output signal.

[0067] The entire system can be understood as a chain of "charge generation - electric field establishment - inductive coupling - charge collection". The intermediate and bottom layers mainly function in the latter three stages.

[0068] Step 1: Charge generation

[0069] Throat vibration → contact separation between the friction layer and the functional layer surface → electron transfer, making the lower surface of the friction layer positively charged (+) and the surface of the functional layer negatively charged (-). These charges are called "tribostatic charges", and they are bound to the insulating polymer surface.

[0070] Step 2: Establishment of the longitudinal electric field

[0071] Because of the accumulation of negative charges on the surface, according to the principles of electrostatics, a strong longitudinal electric field that penetrates the thickness of the film is induced throughout the entire functional layer below and on the bottom electrode.

[0072] Step 3: The core function of the intermediate layer (BST area) – “signal amplifier”

[0073] This increases the intrinsic capacitance of the system, thereby amplifying the output. High-dielectric-constant BST nanoparticles are uniformly dispersed in PDMS, significantly improving the overall dielectric constant of the composite material.

[0074] The entire sensor can be represented as an equivalent capacitor: the upper "plate" is the negatively charged surface layer, the lower plate is the interdigitated electrode at the bottom, and the composite thin film in the middle is the dielectric. The capacitance formula is C=ε. A / d, where ε is the dielectric constant. The addition of BST significantly increases ε, thus significantly increasing the intrinsic capacitance C of the system.

[0075] According to Q=C In the voltage model TENG (where charge = capacitance × voltage), the induced charge Q is relatively constant during vibration (determined by contact electrification efficiency). When the capacitance C increases, the voltage V required to accommodate the same charge Q decreases. However, in TENG's open-circuit voltage model, the more crucial factor is the work done by the charge Q being "forcibly" pulled apart by a distance d during the separation process.

[0076] A more accurate TENG output model shows that high dielectric constant materials can effectively bind charges, reduce charge neutralization, and induce more mirror charges on the electrodes. The ultimate effect is that, under the same mechanical input (vibration), the BST intermediate layer acts as a "charge reservoir" and "signal amplifier," significantly enhancing the amount of transferred charge (ΔQ) and short-circuit current (Isc) that can ultimately be measured in the external circuit. It doesn't generate its own electricity, but rather makes the electricity generated by the top layer "appear" stronger and easier to detect.

[0077] Step 4: The dual role of the bottom layer (HGMs / MWCNTs region) – “Mechanical response regulator” + “charge collection highway”

[0078] HGMs (Mechanical Response Regulators) optimize the mechanical response of sensors to low-frequency, large-amplitude vibrations.

[0079] The incorporation of HGMs reduces the equivalent elastic modulus of the underlying region, making it "softer." When low-frequency, high-amplitude throat vibrations drive the sensor, this "soft" underlying layer allows for greater overall deformation of the entire film. Greater deformation means a larger change in the contact-separation distance (Δd) between the friction layer and the functional layer surface.

[0080] According to the principle of triboelectricity, a larger Δd leads to more intense charge separation and a stronger induced electric field, thereby generating a stronger current / voltage signal in the external circuit. Therefore, by altering local mechanical properties, the underlying layer indirectly but powerfully enhances the sensor's sensitivity to low-frequency signals.

[0081] MWCNTs (charge collection highways) efficiently collect and transfer induced charges to electrodes.

[0082] Although PDMS / BST is an insulator, the negative charge on the surface induces positive charges (mirror charges) on the bottom electrode through the longitudinal electric field. These positive charges need to be quickly and efficiently conducted to the external circuit via the electrodes. MWCNTs form a three-dimensional conductive network at the bottom layer, significantly reducing the interface impedance and transport path resistance of charge flowing from the gradient composite functional layer to the metal electrode. This ensures that the induced charges can be collected almost without loss and converted into a clear electrical signal. Without it, the charge might dissipate within the insulator or respond slowly.

[0083] Step 5: Synergistic Enhancement of Bionic Microcavities

[0084] The resonance of the microcavity amplifies the local mechanical deformation. When the microcavity wall vibrates violently, it disturbs the entire vertical gradient structure around it, enhancing the local contact-separation effect on the upper surface layer and altering the local electric field distribution and capacitance in its depth region (BST layer and below). This violent motion is transmitted through the underlying flexible structure, further optimizing the mechanical response. Therefore, the resonance effect of the microcavity is efficiently converted into an enhancement of the electrical signal by the entire gradient structure.

[0085] Meanwhile, in this embodiment, an artificial laryngeal sensor can be used as the basis for a signal acquisition module for collecting pulse current signals from the interdigital electrode layer, and a signal processing module for calculating laryngeal vibration recognition signals based on the pulse current signals, so as to realize a human-computer interactive voice recognition system.

[0086] Example 2: Sensor fabrication based on gradient centrifugal casting and microcavity depth gradient

[0087] See Figures 1-2 The sensor fabrication process is as follows:

[0088] Solution preparation: PDMS prepolymer and curing agent are mixed at a mass ratio of 10:1 to form the matrix. Three functional slurries are prepared separately:

[0089] Surface slurry A: 0.3 wt% silver nanowires (AgNWs, diameter ~50 nm, length ~10 μm) and 0.1 wt% boron nitride nanosheets (BNNS, thickness ~5 nm) were added to the above PDMS matrix. The mixture was ultrasonically dispersed for 30 minutes.

[0090] Intermediate layer slurry B: 15wt% barium strontium titanate (BST, particle size ~100nm) nanoparticles were added to the PDMS matrix and mechanically stirred for 2 hours.

[0091] Bottom slurry C: Add 8 wt% hollow glass microspheres (HGMs, average diameter ~10 μm, density ~0.6 g / cm³) to the PDMS matrix and stir gently to avoid microsphere breakage.

[0092] Gradient film molding: Slurries C, B, and A are slowly injected sequentially into a cylindrical PTFE mold. The mold is placed in a centrifuge and centrifuged at 3500 rpm for 8 minutes. Under centrifugal force, the higher-density BST particles tend to be evenly distributed, the lower-density HGMs are restricted from floating and mainly distributed at the bottom, while the high aspect ratio AgNWs and BNNS are enriched at the surface (near the air interface) under the combined action of liquid-solid interfacial tension and centrifugal force. Thus, a vertical gradient film precursor with continuous filler transition is formed in one step.

[0093] High aspect ratio structures can form highly efficient functional networks in polymers with extremely low addition amounts. One-dimensional silver nanowires construct long-range conductive pathways, enabling rapid charge collection and transport; two-dimensional boron nitride nanosheets build in-plane thermal conductivity paths and enhance mechanical properties. Together, they impart excellent electrical, thermal, and mechanical enhancement properties to the surface layer with almost no impact on the matrix flexibility, laying the foundation for the sensor's high-frequency response and long-term stability.

[0094] Bionic microcavity array molding: A 3D printing technique was used to fabricate an imprint template with an array of raised micropillars. The height of the micropillars decreased linearly from the center (50 μm) to the edge (20 μm), with a diameter of 30 μm for all pillars. Before the pre-cured film precursor was completely solidified after centrifugation, the imprint template was slowly pressed into the upper surface of the film to a predetermined depth, ensuring that the film material completely encapsulated the micropillars. Subsequently, the film was heated at 80°C for 1 hour to allow the PDMS to initially solidify and set. After demolding, a rectangular or hemispherical blind cavity microcavity array with a depth gradient (from 50 μm in the central region to 20 μm in the edge region) was formed inside the film. This design makes the central region of the sensor more sensitive to low- and mid-frequency vibrations, while the edge region is more sensitive to mid- and high-frequency vibrations, conforming to the spatial distribution of vibration intensity during vocalization.

[0095] Complete curing and electrode fabrication: The gradient film with microcavities was completely cured at 120°C for 2 hours. Gold interdigitated electrodes (IDEs) were fabricated on the bottom surface of the film (i.e., the side enriched with HGMs) using laser direct writing technology. The IDE pattern was designed as three concentric annular interdigitated electrodes (E1 inner ring, E2 middle ring, E3 outer ring). Through optical alignment, the projection areas of E1, E2, and E3 were ensured to correspond to the three annular regions with the deepest, medium, and shallowest microcavities on the film, respectively, achieving regionalized acquisition of charge signals.

[0096] Integration and Encapsulation: A gradient film with IDE (Integrated Device) is transferred onto a flexible polyimide substrate and leads are bonded. A piece of nylon fabric with a microfiber woven on its surface is placed on top of the film as a friction layer. Due to the natural coverage and lack of fixed bonding, friction occurs between the friction layer and the gradient composite functional layer, creating a frictional gap. Finally, biocompatible silicone is used for integrated encapsulation: a breathable medical-grade silicone film is used as the upper layer, a medical-grade acrylic pressure-sensitive adhesive is coated on the lower layer, and release paper is placed on top. The total thickness of the sensor after encapsulation is less than 0.5 mm.

[0097] Example 3: Sensor with micro / nano composite friction layer and fine electrode

[0098] Gradient film preparation: A gradient centrifugation process similar to that in Example 2 was used, but the filler was adjusted: MXene nanoribbons were used to replace part of the AgNWs in the surface slurry; multi-walled carbon nanotubes (MWCNTs) and a small amount of HGMs were used in the bottom slurry to construct a more effective three-dimensional conductive network.

[0099] Microcavity design: Femtosecond laser direct writing technology is used to ablate specific areas of the gradient thin film (corresponding to the future electrode partitions) to form three cylindrical microcavity arrays with diameters of 20μm, 35μm and 50μm, respectively. Their intrinsic frequencies are designed for high frequency, medium frequency and low frequency, respectively.

[0100] Friction Layer Modification: A micro / nano composite structure was fabricated on the surface of an Ecoflex thin film using a secondary template method. First, a micro-pyramid array with a period of 5 μm was replicated on the Ecoflex surface using a micron-scale photoresist template. Subsequently, using this microstructure as a substrate, vertically aligned zinc oxide nanowires (diameter ~100 nm, length ~1 μm) were grown in situ on its surface via a hydrothermal method. This structure significantly increases the contact area and surface roughness, substantially improving the efficiency of triboelectric charge generation, and is particularly sensitive to weak vibrations.

[0101] Electrode pattern correspondence: Using a mask evaporation process, four-quadrant independent interdigitated electrodes precisely aligned with three different sized microcavity array regions are fabricated on the bottom layer of the thin film. The electrode in each quadrant only collects the charge of the microcavity region of the corresponding feature size above it, so that the output signal directly carries spatial-frequency coupling information.

[0102] Achieving a crosslinking gradient: During the preparation of the PDMS slurry, a trace amount of platinum catalyst is added to the surface slurry, while no catalyst is added to the bottom slurry. During the curing process, the surface crosslinking reaction is faster and more complete, resulting in higher crosslinking density and modulus; the bottom crosslinking is slower and less complete, resulting in lower modulus. This crosslinking gradient further enhances the "hard on top, soft on bottom" mechanical gradient, optimizing high-frequency transmission and low-frequency capture capabilities.

[0103] Example 4: Advanced sensor integrating inner wall-modified microcavities and pretreatment circuitry

[0104] Microcavity Inner Wall Modification: Following the microcavity structure formed in Example 2 or 3, an alumina film with a thickness of approximately 10 nm was uniformly deposited on the inner wall of the microcavity using atomic layer deposition (ALD). The dielectric constant of alumina (~9) differs from that of PDMS (~2.7), and this modification layer alters the local electric field distribution and equivalent dielectric constant within the microcavity. When the microcavity deforms due to resonance, the triboelectric effect at the interface between the inner wall alumina layer and the PDMS substrate is enhanced, while the charge binding capacity changes, thereby optimizing the conversion efficiency from mechanical energy to electrical energy and making the resonant amplification effect more pronounced in the electrical signal.

[0105] Integrated signal preprocessing circuit: A miniaturized signal preprocessing circuit module is directly integrated onto a flexible polyimide substrate using flexible electronic printing technology. This module includes:

[0106] Multiplex analog switch: used to sequentially select multiple interdigital electrode channels.

[0107] High input impedance instrumentation amplifier: Differential amplification for each channel to suppress common-mode noise.

[0108] Programmable gain amplifier and anti-aliasing filter: Perform initial conditioning on the signal.

[0109] Ultra-low power microcontroller unit: controls channel switching and can perform simple time-domain feature extraction (such as zero-crossing rate).

[0110] This preprocessing circuit module is directly connected to the lead pads of the interdigital electrodes via flexible conductive adhesive, achieving integrated sensing and processing. It can perform preliminary purification and enhancement before the signal is transmitted over long distances or acquired by the main processor, greatly improving the overall system's anti-interference capability and signal-to-noise ratio.

[0111] In summary, this invention, through ingenious vertical gradient design, biomimetic microcavity resonance, and synergistic innovation of electrodes, packaging, and other systems, has successfully achieved frequency adaptive sensing within a single-layer thin-film structure, providing a core hardware solution for next-generation high-performance, high-comfort, and intelligent wearable artificial larynx and health monitoring devices.

[0112] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A frequency-adaptive triboelectric artificial laryngeal sensor, characterized in that, It includes a friction layer, a gradient composite functional layer, an interdigitated electrode layer, and a flexible substrate arranged from top to bottom; The gradient composite functional layer is a single-layer polymer-based composite film with functional fillers distributed vertically along the thickness direction inside, and a non-penetrating biomimetic microcavity array embedded inside the film; a friction gap is provided between the friction layer and the gradient composite functional layer. The vertical gradient distribution is as follows: the upper surface region near the friction layer is doped with one-dimensional conductive nanomaterials with high aspect ratio and two-dimensional insulating and thermally conductive nanosheets with high aspect ratio; the main body region in the middle of the film is uniformly dispersed with nanoparticles with high dielectric constant; and the lower bottom layer region near the interdigitated electrode layer is doped with low-density microspheres or three-dimensional conductive network materials. The size of the biomimetic microcavity array is designed so that its inherent resonant frequency matches the human speech audio frequency range. The interdigitated electrode layer is fabricated on the lower surface of the gradient composite functional layer; The depth of the microcavity varies in a gradient from the central region of the film to the edge region, in order to correspond to the spatial distribution differences of different vibration modes on the throat surface. The pattern of the interdigitated electrode layer consists of multiple concentric ring interdigitated electrodes that are electrically insulated from each other, and the projection position of each ring interdigitated electrode on the plane corresponds to the microcavity regions of different sizes or different distribution densities in the biomimetic microcavity array. The gradient composite functional layer is a polydimethylsiloxane with a gradient change in crosslinking degree, wherein the crosslinking degree of the upper surface region is higher than that of the lower bottom layer region; The inner wall of the biomimetic microcavity is modified with a nano-coating with a dielectric constant different from that of the substrate material.

2. The artificial laryngeal sensor according to claim 1, characterized in that, In the biomimetic microcavity array, the shape of the microcavity is at least one of cylindrical, conical, or hemispherical.

3. The artificial laryngeal sensor according to claim 1 or 2, characterized in that, In the vertical gradient distribution, the mass fraction of the one-dimensional conductive nanomaterial in the upper surface region is 0.05%-0.5%, the mass fraction of the two-dimensional insulating and thermally conductive nanosheet is 0.01%-0.2%, the mass fraction of the high dielectric constant nanoparticles in the middle main region is 5%-20%, and the mass fraction of the low-density microspheres or three-dimensional conductive network material in the lower bottom layer region is 2%-10%.

4. The artificial laryngeal sensor according to claim 1, characterized in that, It also includes an integrated flexible encapsulation layer, which completely encapsulates the friction layer, the gradient composite functional layer, the interdigitated electrode layer and the flexible substrate. The encapsulation portion above the friction layer is a stretchable breathable film, and the encapsulation portion below the flexible substrate is a biocompatible pressure-sensitive adhesive layer.

5. The artificial laryngeal sensor according to claim 1, characterized in that, The friction layer is a polymer film with a micro-nano composite structure on its surface. The micro-nano composite structure includes micron-scale protrusions formed by a template method and nanowires or nanoparticles grown in situ on the surface of the protrusions.

6. The artificial laryngeal sensor according to claim 1, characterized in that, The vertical gradient distribution is formed in one step by gradient centrifugal casting, in which different fillers spontaneously form a continuous gradient distribution along the thickness direction under the centrifugal force field due to density differences.

7. A human-computer interaction speech recognition system, characterized in that, The system includes: Artificial laryngeal sensor as described in any one of claims 1-6; The signal acquisition module is used to acquire the pulse current signal of the interdigital electrode layer; The signal processing module is used to calculate the throat vibration identification signal based on the pulse current signal.

Images