A soft, flexible and communicative device for patients with laryngectomy and its method thereof

Abstract

The present invention relates to a soft, flexible, and communicative device for patients with laryngectomy and its method thereof The present invention discloses the fabrication of soft, lightweight, skin-compatible, and reusable device for silent speech interfaces (SSIs). It further comprises the unique feature to recognize speech-related information using surface electromyography (sEMG). sEMG-based speech recognition operates on signals recorded from a set of sEMG sensors that are strategically located on the face (with the help of desired form factor) and measure muscle activity associated with the phonation, resonation, and articulation of speech. sEMG sensors are printed onto strategical locations on soft, stretchable, and biocompatible membrane shaped into a form-factor of a face mask. The silent speech interfaces (SSIs) can also be deployed in acoustically challenging environment or where privacy / confidentially is a desirable such as in defense or military application and it is cost effective with reliable speech recognition accuracy of 96.2%(>95 %).

Description

[0001] A SOFT, FLEXIBLE AND COMMUNICATIVE DEVICE FOR PATIENTS WITH LARYNGECTOMY AND ITS METHOD THEREOF

[0002] FIELD OF INVENTION

[0003] The present invention relates to a device for patients with laryngectomy and its method thereof. More specifically, the present invention relates to a method to provide soft, flexible and communicative device aiming to recognize speech-related information using surface electromyography (sEMG), for patients who have a laryngectomy and often struggle with their daily communication. The present invention further relates to the fabrication of soft, lightweight, skin-compatible and reusable devices for silent speech interfaces (SSIs).

[0004] BACKGROUND OF THE INVENTION

[0005] Human speech is a natural and efficient means of communication, yet millions of people around the world with severe speech disorders are unable to communicate effectively through vocalization. Instead, depending on the nature of the speech disorder, they must rely on augmentative and alternative communication (AAC) devices or software. These include artificial voice sources after loss of laryngeal function and / or speech synthesizers for individuals unable to articulate speech sounds. Unfortunately, these alternative communication solutions typically provide unnatural sounding vocalization or require the involvement of the user’s hands, thus complicating everyday interactions and making them unwieldy.

[0006] One technology that has been leveraged for assisting those with speech disorders is automatic speech recognition (ASR), in which acoustic speech is translated into a sequence of speech tokens, typically words, using pattern classification techniques. ASR performance for those with normal speech function has achieved accuracies approaching 100%, permitting effective commercial applications and integration into portable speech-based human-machine interfaces. However, as successful as ASR has been for the general population, research on ASR of disordered speech is limited and has almost exclusively focused on recognition of acoustic speech. Individuals who have lost the ability to speak normally cannot make full use of ASR interfaces, even if their language function is intact. ASR performance also degrades rapidly in the presence of acoustic noise, rendering it unsuitable for use in acoustically harsh environments, and it lacks privacy when used as a computer interface.

[0007] Each year thousands of individuals require surgical removal of the larynx (voice box) due to trauma or disease, and thereby require an alternative voice source or assistive device to verbally communicate. Although natural voice is lost after laryngectomy, most muscles controlling speech articulation remain intact. Surface electromyographic (sEMG) activity of speech musculature can be recorded from the strategic locations of the face and used for automatic speech recognition to provide speech-to-text or synthesized speech as an alternative means of communication. This is true even when speech is mouthed or spoken in a silent (sub vocal) manner, making it an appropriate communication platform after laryngectomy.

[0008] There are many methods available for these individuals known in the art for patients with laryngectomy. For example, W01994001059Alrelates to an artificial larynx comprising of a first unit mounted in the mouth comprising a dental prosthesis including a loud speaker, a power amplifier, a self-contained power source, plus a radio frequency receiver, and a second unit held in the hand equipped with an input control device, self-contained power source, electronic circuitry and transmitter allowing the user to alter the frequency and volume produced by the unit in the mouth.

[0009] US4502151A relates to a Replaceable saliva barrier for intra-oral larynx where an intra-oral artificial larynx is provided including a signal generator and a speaker mounted on prosthesis for mounting the larynx within the intra-oral cavity. A horn is provided by having an input opening coupled to the speaker for acoustically amplifying the output of the speaker, the horn also having an output opening. A saliva barrier is removably mounted to the output opening of the horn for precluding penetration of saliva into the horn, the barrier having a first end which is telescopically received within the output opening of the horn and the second end having a mouth formed on a bias to the longitudinal axis of the mounting and a microporous membrane sealing the mouth. The mouth of the mounting has an elliptical shape which has been found to increase speaker volume. Moreover, the telescopic mating of the mounting and the horn provides a saliva barrier which is easily replaceable by the user.

[0010] However, there are currently only a limited number of post-laryngectomy voice restoration methods available for these individuals and unfortunately, these methods are often limited by their usability and / or the abnormal voice produced, which may be hard to understand for listeners. Although, emerging assistive technologies (ATs) such as silent speech interfaces (SSIs) have shown promising potential in recent years as an alternate solution.

[0011] SSIs are devices that enable speech communication to take place in the absence of audible acoustic signals. To date, a number of SSIs have been proposed in an attempt to extract non-acoustic information generated during speech production and reproduce audible speech using different sensing modalities, such as measuring electrical activities of the brain or the articulator muscles, or by capturing movements of the speech articulators themselves.

[0012] There is, however, a need for improvement. The methods used over a long period of time have resulted in endurance. Despite the attractive attributes of SSIs, there are still challenges in the form of hardware (e.g. portability, lightweight, unobtrusiveness and wearability) and processing software (e.g. efficiency, robustness, and intelligibility speech generation). Therefore, there is a need for the fabrication of soft, lightweight, skin-compatible, and reusable devices for SSIs. Moreover, there is a need to provide a soft, flexible and communicative device for patients with laryngectomy to recognize speech-related information using surface electromyography (sEMG)which overcomes some of the existing problems and can be made easily without much complex manufacturing method or process. The present inventors have surprisingly developed an effective device which ameliorates the aforesaid shortcomings of the prior art. In addition to that, the present device is robust, efficient in speech generation (accuracy >95 %) as well as economically affordable to the needy persons.

[0013] OBJECT OF THE INVENTION

[0014] The principal object of the present invention is to provide a soft, flexible and communicative device for patients with laryngectomy.

[0015] Another object of the present invention is to provide a soft, flexible, and communicative device and its method for post-laryngectomy voice restoration.

[0016] Another object of the present invention is to provide fabrication of soft, lightweight, skin-compatible and reusable device for silent speech interfaces (SSIs).

[0017] Another object of the present invention is to provide patients who have a laryngectomy with a flexible communicative device having emerging assistive technologies (ATs) such as silent speech interfaces (SSIs) providing promising potential.

[0018] Yet another object of the present invention is to provide fabrication and optimization of surface electromyography (sEMG) sensors on a suitable substrate being used for mask (either textile or a transparent breathable polymer).

[0019] Further object of the present invention is to provide a device with silent speech interfaces (SSIs) that enable speech communication to take place in the absence of audible acoustic signals.

[0020] Further object of the present invention is to provide a device with monopolar configuration of EMG for the present module saves the number of electrodes to be used and adds to the convenience of the user.

[0021] Yet another object of the present invention is to provide a device which accounts for minimum power dissipation and crosstalk.

[0022] Further object of the present invention is to provide a device which can be used as a wearable device alternative for a 4 channel EMG device.

[0023] Yet another object of the present invention is to provide a device with the Printed circuit board (PCB) designed in such a way that two channels on one side are connected to another two channels with the help of serpentine interconnects along with the desirable components.

[0024] Further object of the present invention is to provide a device with embedded electronic components into a neckband for wireless detection.

[0025] Yet another object of the present invention is to provide a device aimed at processing surface electromyography (sEMG) signals for speech recognition.

[0026] Further object of the present invention is to provide a device, which is robust, efficient in speech generation (accuracy >95 %) as well as economically affordable to the needy persons.

[0027] Further object of the present invention is to provide a soft, flexible, and communicative device, which is user friendly, wireless, reusable, and safe.

[0028] SUMMARY OF THE INVENTION

[0029] The present invention relates to a soft, flexible and communicative device for patients with laryngectomy and its method thereof. The present invention discloses the fabrication of soft, lightweight, skin-compatible and reusable device for silent speech interfaces (SSIs). The silent speech interfaces (SSIs) have been proposed to extract non-acoustic information generated during speech production and reproduce audible speech using different sensing modalities, such as measuring activities of the brain or the articulator muscles, or by capturing movements of the speech articulators themselves. The soft and flexible device further comprises the unique feature to recognize speech-related information using surface electromyography (sEMG). sEMG-based speech recognition operates on signals recorded from a set of sEMG sensors that are strategically located on the face (with the help of desired form factor) and measure muscle activity associated with the phonation, resonation, and articulation of speech. sEMG sensors are printed onto the strategical locations on a soft, stretchable and a biocompatible membrane shaped into a form-factor of a face mask. The silent speech interfaces (SSIs) can also be deployed in acoustically challenging environment or where privacy / confidentially is desirable (such as in defence or military applications), and not limited to its use as a communication aid for speech impaired individuals. Brief Description of Drawings:

[0030] Other objects, advantages and novel features of the invention will become apparent from the following detailed description of the present embodiment when taken in conjunction with the accompanying drawings.

[0031] Figure 1 (a) shows the schematic diagram for the fabricated facemask exhibiting transparency, breathability and antifogging properties; along with the (b) photographic images of the developed prototype.

[0032] Figure 2 depicts the block diagram for the electronics module to be embedded into the neck band of the mask.

[0033] Figure 3 describes the block diagram for the signal processing steps to be incorporated for synthesizing speech from sEMG signals.

[0034] Figure 4 describes the block diagram outlining the preparation process of the membrane.

[0035] Figure 5 (a) demonstrating the membrane test for waterproofing (b) demonstrating transparency of the membrane (c) demonstrating antifogging property of membrane (d) demonstrating breathable property of membrane.

[0036] Figure 6 depicts the (a) Schematic of the screen-printing process, (b) rheological properties of the synthesized ink, (c) electrical properties of the ink after screen printing (multiple layers) and thermal annealing, (d) schematic illustrating the anatomical locations of the sensors, and (e) demonstration of the final prototype.

[0037] Figure 7 describes the (a) Schematic showing the electrodes on the facial muscles, where 1, 2, 3 4, R and G designate the channels 1, 2, 3, 4, reference and ground electrodes respectively, (b) 4 Channel PCB connection block diagram for Data extraction (c) The connection of 4 Channel PCB with ESP32 for Data extraction over Wi-Fi. Figure 8 depicts the modules for extraction of facial expression data to wearable devices as shown in fig. 7.

[0038] Figure 9(a) depicts data extracted and corresponding FFT from Seed Studio module for the word 'Graphene' (b) Data extracted and corresponding FFT from 4 channel

[0039] 5 PCB for the word 'Graphene'.

[0040] Figure 10 describes schematic diagram for 4 channel PCB module.

[0041] Figure 11 and 12 describes the circuit and corresponding bode plot respectively.

[0042] Detailed Description of Invention:

[0043] Before explaining the present invention in detail, it is to be understood that the0 invention is not limited in its application to the details of the mechanism and arrangement of parts illustrated in the accompanying drawings. The invention is capable of other embodiments, as depicted in different figures as described above and of being practiced or carried out in a variety of ways. It is to be understood that the phraseology and terminology employed herein is for the purpose of description and not of limitation.

[0044] The present invention is a soft, flexible and communicative device for patients with laryngectomy. It is more related to the fabrication of soft, lightweight, skincompatible and reusable device for emerging assistive technologies (ATs) such as silent speech interfaces (SSIs) along with its method. SSIs are devices that enable0 speech communication to take place in the absence of audible acoustic signals.

[0045] The present invention further provides a method to fabricate a flexible device which demonstrates the potential of using a surface electromyography (sEMG) based silent speech recognition system as the basis of an alternative communication device for persons living with laryngectomy. The method comprising of: f sEMG sensors are fabricated on a suitable substrate used for mask either textile or a transparent breathable polymer using conducting material such as metal (gold, silver or copper), carbon, graphene or a polymer such as PEDOT:PSS [poly(3,4- ethylenedioxythiophene):poly(styrenesulfonate)] as an electrode material, b. The Patterns comprising sensor area as well as serpentine inter connects are screen printed on to the plasma cleaned substrate, so as to ensure a better adhesion, c5 Sensors are connected further to the data acquisition module (miniaturized printed circuit board with chips such as AD620, LM358) through snap buttons / connectors for the transduction of the signals, d. The sensor locations are selected so as to get the signals from the articulator muscles of the face during speech tasks, k£) The signals are finally collected and processed for speech recognition by varying vocabulary sizes (15 words to 1000 words) and using different algorithms and Artificial Neural Network (ANN).

[0046] Example Figs. 1, 2 and 3 further indicate the preliminary work been performed in this direction to demonstrate the proof of concept.

[0047] 15 Further, the locations for the electrodes for signal transduction are decided on the basis for the facial muscles involved specifically in speech generation.

[0048] A transparent, breathable and waterproof membrane has been employed as a platform for embedding the sEMG sensors. The proper adhesion of conducting material with this membrane (after surface treatment) allows it to be reused after 0 cleaning with soap solution or sanitizer without any significant loss of sEMG signals.

[0049] The electrodes are printed onto the desired substrate locations using screen printing. The ink material and the said process are optimized to get the desired biosignals. sEMG signals thus extracted are transduced and processed for speech recognition.

[0050] The present invention is a communicative device, aiming to recognize speech- 5 related information using surface electromyography (sEMG). sEMG-based speech recognition operates on signals recorded from a set of sEMG sensors that are strategically located on the face (with the help of desired form factor) and measure muscle activity associated with the phonation, resonation and articulation of speech. sEMG sensors are printed onto the strategical locations on a soft, stretchable and a biocompatible membrane shaped into a form-factor of a face mask.

[0051] The present invention encompasses a communicative device which primarily derives three main parts, although it is not limited to these aspects alone. The device is designed to enhance communication capabilities through a combination of innovative materials, embedded technologies, and advanced algorithms.

[0052] The first part of the invention (A) focuses on the creation of a face mask made from a unique, flexible, and transparent membrane that is both breathable and waterproof. (B) Embedded within the mask are strategically placed surface Electromyography (sEMG) sensors, which are designed to detect and record electrical signals generated by facial muscles during speech. These sensors are carefully integrated into the mask in such a way that they do not interfere with its functionality or wearability. The process of embedding the sensors involves the fabrication of the mask material and attaching the boundary material using ultrasonic welding technique that ensures durability, flexibility, and seamless integration of the sensors without compromising the integrity of the mask's design. The mask’s transparent nature allows for clear visibility of the user's facial expressions while remaining comfortable and functional. The mask as prepared in the present invention comprises of a transparent membrane at the portions over the nose and mouth of the wearer.

[0053] In one aspect of the invention, i.e. (A), the membrane is formulated in such a way that it allows the water vapors to permeate through it while blocking the water droplets to pass through. Water vapor transmission is achieved while maintaining optical transparency. Further, the inherent hydrophobic nature of the membrane does not allow the microbes or infectious bacteria to adhere on its surface thus avoiding the use of any antimicrobial coatings such as Organosilane Quaternary Amino which otherwise can also be harmful to the wearer and pose environmental challenges at the same time.

[0054] In another aspect of the invention, the mask material is coated with an antifogging solution to avoid the fogging of the material while breathing or talking. The antifogging method presented here is comparatively simpler than the one used in prior art.

[0055] Further the membrane incorporated in the present invention is fabricated by using a Polymeric Organo Silicone Compound (POSC) such as Poly dimethyl siloxane,

[0056] 5 Octaphenylcyclotetrasiloxane, Heptamethyltrisiloxane, which is further manipulated to have morphology and internal structure so as to facilitate breathability while maintaining optical transparency and hydrophobicity. POSC used here is a two- component system including a prepolymer and a cross-linking agent (CA).

[0057] The method for membrane preparation comprising of following steps: lOi. mixture of prepolymer and a lower boiling point solvent (such as chloroform, dimethyl formamide, xylene, ethyl acetate or a mixture thereof) in the ratio 1 : 1 followed by mixing at 400-500 rpm for 5-10 minutes. ii. a higher boiling point nonsolvent (water, ethanol, isopropanol, n-propyl alcohol or a mixture thereof) is introduced dropwise into the above mixture in the ratio of 1 :2

[0058] 15 (prepolymer to non-solvent). iii. The mixture is stirred at 400-500 rpm for 10-15 minutes for complete mixing. iv. The cross-linking agent is further added into the mixture in the ratio of 10: 1 (prepolymer to cross-linking agent). v.The final mixture is vacuum degassed to remove any air bubbles for 30-40 minutes. 0vi. After degassing, the mixture is cast into a mold with the defined spacers to have desired thickness of the membrane. vii. The mixture is allowed to dry at room temperature for 15-20 minutes to remove the lower boiling point solvent and then heated at 120°C for 10 minutes to evaporate the higher boiling point solvent and cross-linking of POSC simultaneously. 5 The cross-linked membrane is then procured from the mold and cut into the desired size and shape. The membrane is further coated with an antifogging solution followed by a surface treatment. After blotting the excess solution and drying the membrane, final resulting transparent, breathable, waterproof and antifogging membrane is obtained. The brief schematic diagram illustrating the methodology is shown in Fig 4. The membrane is further tested for its breathability, transparency, waterproof and antifogging properties. As shown in Fig. 5 the hydrophobic, and hence waterproof, property of the membrane is depicted by water contact angle value of 111.2° ± 2°, that is, greater than 90° (Fig. 5a).

[0059] Further for demonstrating transparency of the membrane, it is placed over the logo of “CSIR-CSIO” printed on a paper (Fig. 5b).

[0060] The antifogging property has been illustrated by placing the membrane on the opening of a bottle containing boiled water for 1-2 minutes. For comparison, the same experiment is performed with the membrane not treated with the antifogging solution. The untreated membrane is found to be completely fogged, whereas the treated membrane is completely clear and free of any fog (as shown in Fig. 5c).

[0061] The breathability of the membrane has been confirmed from the Water Vapour Transmission Rate (WVTR) values (Table 5d), which are even higher (0.682 g / m2 / h) for the fabricated membrane than a filter paper (0.577 g / m2 / h).

[0062] The membrane thus formulated is used for preparing face mask by attaching a boundary material such as cloth by using ultrasonic welding. The whole membrane can be used at the boundary portion as well as at portions of nose and mouth for full visibility. Further, the elastic loops are attached at both ends of the mask material with nose wire at the top center.

[0063] Further another aspect of the invention 1 (B) is embedded within the mask with strategically placed surface Electromyography (sEMG) sensors and formulation of the same to the shape of facemask which attaches to the boundary material with the help of ultrasonic welding.

[0064] As shown in Fig. 6, for sensors fabrication, the screen printing process is used where, screen of desirable mesh size is patterned in the required design and ink is allowed to flow through the patterned screen with the help of a squeeze (as shown in figure 6a), As shown in Fig. 6b, for sEMG sensors, conducting ink such as graphene is employed and it is optimized for rheological properties so as to be employed for screen printing process.

[0065] As shown in Fig. 6c, before printing the membrane material is subjected to surface treatment with the help of plasma oxidation to improve the adhesion strength of the printed material to the membrane substrate. The sheet resistance is measured for the printed multiple layers with the subsequent step of annealing, where the resistance values get decreased with the increase in number of printed layers.

[0066] As shown in Fig. 6d, the design of the printed pattern is selected to have the sensors onto the strategic locations of the facial muscles associated with the speech process.

[0067] And then the membrane is formulated into the shape of the mask by attaching the boundary materials with the help of ultrasonic welding. The final prototype image is shown in Fig. 6(e).

[0068] The second part of the invention is focused on the fabrication of the module for extracting sEMG signals and wireless transmission. It is focused on the development of a specialized module responsible for extracting the sEMG signals detected by the sensors. This module is designed to process the electrical signals generated by the facial muscles, filter out noise, and convert the signals into a usable format. Once processed, the module is equipped with a wireless transmission system that enables the signals to be transmitted to an external device, such as a smartphone or a computer, in real-time.

[0069] It provides an EMG Data Acquisition module for accurate and consistent data for all the 4 channels using monopolar EMG configuration. It can distinguish EMG signals corresponding to different facial expressions, which requires better Signal to Noise Ratio as well as optimized gain. It is further formed as the compact system which can be used as wearable device, all the 4 channels are on the same Printed Circuit Board (PCB) and are able to run on single power supply with 3.3 V power source.

[0070] The process begins (as referred to fig. 8) with a module component for implementation of data extraction to corresponding system which can be used as a wearable device. The sEMG Data acquisition module for wearable devices involves capturing the steps involved from data acquisition to transmission.

[0071] As shown in Fig. 7a, the electrode position is shown for extracting facial EMG data. For data extraction with the 4 Channel PCB wireless data transmission is more convenient. For this purpose, a standard ESP 32 development board ESP WROOM 32' is used which supports Wi-Fi. The data was transmitted to the PC over Wi-Fi at the sample rate of 1000 and baud rate 115200. The ESP 32 board was powered from USB cable and the 3.3 V pin of it was used to power the 4 Channel PCB. The GPIO pins 34,35, 32 and 33 related to the output of first, second, third and fourth channel of 4 Channel PCB respectively. The block diagram and the image of the setup used for data extraction is shown in Figure 7b and 7c respectively. The optimum dimension for the current PCB was found to be 40 by 50 mm, which works best without any crosstalk or external noise interference.

[0072] The proposed 4 channel PCB is able to get data for facial expression reliably for all the four channels, with the size of the PCB was optimized as per the use of wearable applications. The comparison plot for the signal and its FFT are shown in figure 9a (Seed Studio module) and figure 9b (4 Channel PCB). The EMG data was extracted while speaking the word ‘Graphene’.

[0073] Further as shown in Fig. 10, the 4 channel EMG module is designed in the form of a 2-layer PCB having dimensions as 40 X 50 mm. The PCB was designed in Ki-Cad software, the schematic for the same is shown in figure 10.

[0074] The further working of the 4-channel setup can be described as the circuit is designed to work on the power supply of as minimum as 3.3V. As this is a single supply circuit the reference voltage for the AD623 and LM324 is provided using a potential divider circuit which supplies half the supply voltage as reference as shown in figure 10. For potential divider an IC OPA333 was used as shown in figure 19. There are two 10 k resistor and one capacitor of 100 nF which goes into the positive pin of OPA333 which is then given unity feedback to obtain 1.65 volt as the output. Thus, the output signal of the circuit oscillates around 1.65 volts. For each channel the instrumentation amplifier AD623 provides a gain of 500. The value of gain resistor is taken as 200 ohm as per the datasheet. Each Op-amp of LM324 is used for a bandpass filter. As per the Bode plot the maximum gain of band pass filter is -IdB which corresponds to the linear gain of 0.9. The circuit and corresponding bode plot as referred to figure 11 and 12 respectively. So, the overall gain of each channel is the product of the gain of both stages which is 450 approximately. This gives a 4-channel module which amplifies the EMG with a gain of 450 within a frequency range of 15- 150 Hz and the output signal oscillating around 1.65 volts.

[0075] The final part of the invention involves the development capable of converting sEMG signals into corresponding speech / audio. In order to further process the extracted sEMG signals into speech, the data was collected at a sampling rate of 1000 Hz. To minimize noise interference, the collected data was filtered using a bandpass filter with a range of 20-450 Hz. The filtered data was then divided into 3- second segments for feature extraction. Time-domain features were extracted from each segment, and these features were used to train a machine learning model. The extreme Gradient Boosting (XGBoost) classifier was employed for the classification task. The dataset was split into 90% for training and 10% for testing. After training, the model was used for speech conversion, with the accuracy for converting to speech corresponding to different letters / words reaching 96.2%

[0076] Further, the present invention is not limited to the form-factor of face mask, it can be modified as per the requirements such as a band or ear cap of earbuds for measuring sEMG or other biopotential signals such as EEG for SSIs.

[0077] Moreover, SSIs can also be deployed in acoustically challenging environments or where privacy / confidentially is desirable (such as in defense or military applications) and not limited to its use as a communication aid for speech impaired individuals. The present device can be used for long-term applications, which otherwise is lacking in the existing devices available. The wireless functionality of the device eliminates the need for cumbersome wires and enhances the device's usability in various environments, allowing for more convenient communication for individuals with speech impairments or other communication challenges. Moreover, the reduced cost of the product, mainly because of the simplicity of the process and inexpensive materials employed, is an asset here. Further the device is robust, efficient in speech generation (accuracy >95 %) as well as economically affordable to the needy persons. In addition to that the device can be reused multiple times and for a longer duration as the mask material is washable and embedded electrodes retain their properties even after several cycles of washing.

Claims

AMENDED CLAIMS received by the International Bureau on 08 October 2025 (08.10.2025)We Claim:

1. A soft, flexible and communicative device for patients with laryngectomy, comprising: characterized by a transparent, flexible, and breathable membrane that is waterproof and covers the portions over the nose and mouth of the wearer; embedded surface Electromyography (sEMG) sensors placed within the mask to detect and record electrical signals generated by facial muscles during speech; the sEMG sensors being integrated into the mask without interfering with its functionality or wearability, such that they remain comfortably positioned on the wearer's face; embedding the sensors within the mask material by ultrasonic welding, which ensures the sensors durability, flexibility, and seamless integration.

2. The soft, flexible and communicative device for patients with laryngectomy as claimed in claim 1, wherein the ultrasonic welding technique integrated for attaching the mask material, containing sensor, to the boundary material ensures a durable, flexible, and waterproof interface between the mask material and the boundary material.

3. The soft, flexible and communicative device for patients with laryngectomy as claimed in claim 1, wherein extracting sEMG signals from the face mask is embedded with a module designed to process the electrical signals detected by the sEMG sensors.

4. The device as claimed in claim 1, wherein the sEMG sensors are integrated into an alternative form factor selected from the group consisting of a band, an ear cap of earbuds, or other wearable structures, to facilitate the measurement of sEMG or other biopotential signals such as EEG for SSIs.

5. A method to provide a soft, flexible and communicative device for patients with laryngectomy, comprising; a. fabricating sEMG sensors on a suitable substrate for the mask, which can be either textile or a transparent breathable polymer, using a conducting material;b. patterning a sensor area and serpentine interconnects using screen printing process onto the plasma-cleaned substrate, ensuring improved adhesion of the sensor patterns to the substrate; c. connecting the sensors to a data acquisition module, wherein the module is a miniaturized printed circuit board (PCB) with chips such as AD620 or LM358, through snap buttons or connectors to facilitate the transduction of electrical signals generated by the sensors; d. selecting sensor locations on the mask to capture signals from the articulator muscles of the face during speech tasks, ensuring accurate signal capture for speech recognition; e. collecting and processing the sEMG signals for speech recognition, wherein the collected signals are used with varying vocabulary sizes ranging from 15 words to 1000 words, and processed using different algorithms and an Artificial Neural Network (ANN) for speech recognition.

6. The method as claimed in claim 5, wherein the conducting material in step (a) is selected from the group consisting of metal for instance gold, silver, copper, carbon, or graphene.

7. The method as claimed in claim 5, wherein the polymer in step (a) is selected as PEDOT:PSS [poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate)] as the electrode material.

8. The method as claimed in claim 5, wherein the surface Electromyography (sEMG) signals converted into speech recognition, comprising:I. placing a face mask on the wearer, wherein the mask is made of a transparent, flexible, and breathable membrane that covers the portions over the nose and mouth and is waterproof;II. embedding surface Electromyography (sEMG) sensors within the mask, wherein the sensors are positioned to detect electrical signals generated by the facial muscles during speech without interfering with the wearability and functionality of the mask;III. ultrasonically welding the mask material containing sEMG sensors to the boundary material so as to ensure durability, flexibility, and seamless integration without compromising the mask's design;IV. collecting sEMG signals from the embedded sensors during facial muscle movements, corresponding to speech production, at a sampling rate of 1000 Hz;V. filtering the collected sEMG signals using a bandpass filter with a frequency range of 0-400 Hz to minimize noise interference from external sources;VI. segmenting the filtered signals into 3-second segments for feature extraction;VII. extracting time-domain features from each of the 3-second segments of the filtered sEMG signals;VIII. training a machine learning model using the extracted time-domain features, wherein an extreme Gradient Boosting (XGBoost) classifier is employed to classify the features corresponding to distinct speech patterns;IX. splitting the dataset into 90% for training and 10% for testing to validate the performance of the trained model;X. converting the processed sEMG signals into corresponding speech or audio output; andXI. wirelessly transmitting the processed sEMG data from the mask’s embedded module to an external device for real-time speech conversion.

9. The method as claimed in claim 8, wherein the trained machine learning model achieves an accuracy of 96.2% for converting the signals into distinct letters or words.

Images