Speech Recognition Applications Research Articles

Intelligent Clinical Documentation: Harnessing Generative AI for Patient-Centric Clinical Note Generation

Comprehensive clinical documentation is crucial for effective healthcare delivery, yet it poses a significant burden on healthcare professionals, leading to burnout, increased medical errors, and compromised patient safety. This paper explores the potential of generative AI (Artificial Intelligence) to streamline the clinical documentation process, specifically focusing on generating SOAP (Subjective, Objective, Assessment, Plan) and BIRP (Behavior, Intervention, Response, Plan) notes. We present a case study demonstrating the application of natural language processing (NLP) and automatic speech recognition (ASR) technologies to transcribe patient-clinician interactions, coupled with advanced prompting techniques to generate draft clinical notes using large language models (LLMs). The study highlights the benefits of this approach, including time savings, improved documentation quality, and enhanced patient-centered care. Additionally, we discuss ethical considerations, such as maintaining patient confidentiality and addressing model biases, underscoring the need for responsible deployment of generative AI in healthcare settings. The findings suggest that generative AI has the potential to revolutionize clinical documentation practices, alleviating administrative burdens and enabling healthcare professionals to focus more on direct patient care.

Speech Emotion Recognition

Speech emotion recognition is a vital area of research with applications starting from human-computer interaction to mental health monitoring. This paper will provide a comprehensive survey of the techniques, methods, applications, and challenges in speech emotion recognition. It begins by checking the significance of recognizing emotions from speech and its diverse applications across various fields. The paper then shows the method employed for emotion speech recognition, encompassing traditional machine learning techniques, such as support vector machines and Gaussian mixture models, as well as contemporary approaches, including deep learning and multimodal fusion. Moreover, it examines benchmark datasets commonly used for training and evaluation purposes in emotion speech recognition research. Speech Emotion Recognition (SER) has a wide range of applications and there has been a lot of research going on in this fascinating area in recent years. However, the entertainment sector suffers from a lack of study in this research. Many of use Neural Network (NN) and Long Short-Term Memory (LSTM) architectures to categorize the emotions in audio recordings captured by actors expressing various emotions. Moreover, our survey explores real-world applications of emotion speech recognition like virtual assistants, health sector, market sector, education, Mental health diagnosis. The paper discusses the challenge associated with emotion speech recognition, including the variability of emotional expressions, cultural influences, humor and privacy concerns. In future it will help others to deal with many Noisy dataset and various cultural effects of emotion.

Wake Word and Speech Recognition Application on Edge Device: A Case of Improving the Electric Wheelchair

Wake Word and Speech Recognition Application on Edge Device: A Case of Improving the Electric Wheelchair

A Novel Approach for Realizing Loop-Filter in Low-Distortion Noise-Coupled ΣΔ Modulators

This paper presents an innovative general structure for a single-loop Sigma-Delta ([Formula: see text]) modulator that combines low-distortion and noise-coupled techniques to achieve high-resolution for low-power applications. The low-distortion technique adjusts the signal transfer function of the proposed structure to unity, while the noise-coupled technique enhances the order of quantization noise-shaping at the modulator output. These techniques were incorporated into the design to increase the modulator order without requiring additional operational amplifiers during its circuit implementation, resulting in a low-power modulator compared to similar structures. To reduce the number of required amplifiers, a second-order infinite impulse response (IIR) filter was used in the modulator loop filter, in place of two integrators. To assess the proposed structure’s performance, a third-order modulator sample was implemented and simulated for speech recognition applications, specifically, digital hearing aids, using 180[Formula: see text]nm complementary metal oxide semiconductor (CMOS) technology. The results indicate that for a third structure with a sampling rate of 64, an input sine signal of −6[Formula: see text]dBFS and a sampling frequency of 2.56[Formula: see text]MHz, the signal to noise plus distortion (SNDR) was 81.9[Formula: see text]dB, and the dynamic range (DR) was 88[Formula: see text]dB for a 20[Formula: see text]KHz bandwidth. The modulator’s power consumption was 126.9[Formula: see text][Formula: see text]W. Circuit and system-level simulations demonstrate the effectiveness of the proposed method.

Highly-performance nonwoven pressure sensors with enhanced breathability and durability for human health monitoring

In recent years, the rapid development of smart wearable devices has resulted in significant advances in flexible pressure sensors, which have found application in speech recognition, health monitoring, as well as motion tracking. There is, however, still a significant challenge associated with the low-cost preparation of flexible pressure sensors with high sensitivity and permeability. An innovative nonwoven-based flexible pressure sensor is presented in this paper consisting of a conductive sensing layer composed of reduced graphene oxide-treated polyester hydrophilic nonwoven, an electrode layer comprised of carbon nanotube-printed polyester plain nonwoven electrodes, and a top pressure layer composed of hydroentangled viscose dome structure nonwoven. The pressure sensor exhibits high sensitivity (5.43% kPa−1 in the range of 0–10.12 kPa), excellent permeability (807.2 mm s−1), rapid response and recovery time (73/73 ms), a wide sensing range (0–185 kPa), and remarkable cycling durability (∼20,000 cycles). The sensor exhibits superhydrophobic properties, as indicated by a water contact angle of up to 155°. Based on its commendable performance, the sensor demonstrates the capability to discern various pressure signals, enabling the potential for encrypted transmission of military information. Additionally, it proves valuable in applications such as health monitoring and motion tracking. Consequently, the flexible pressure sensor has significant potential for application in areas such as data encryption, health monitoring, and motion detection.

SPEECH RECOGNITION SYSTEM

Speech recognition technology has witnessed remarkable progress in recent years, fueled by advancements in machine learning, deep neural networks, and signal processing techniques. This paper presents a comprehensive review of the current state-of-the-art in speech recognition systems, highlighting key methodologies and breakthroughs that have contributed to their improved performance. The paper explores various aspects, including acoustic modeling, language modeling, and the integration of contextual information, shedding light on the challenges faced and innovative solutions proposed in the field. Furthermore, the paper discusses the impact of large-scale datasets and transfer learning on the robustness and adaptability of speech recognition models. It delves into recent developments in end-to-end models and their potential to simplify the architecture while enhancing accuracy. The integration of real-time and edge computing for speech recognition applications is also explored, emphasizing the implications for practical implementations in diverse domains such as healthcare, telecommunications, and smart devices. In addition to reviewing the current landscape, the paper provides insights into future prospects and emerging trends in speech recognition research. The role of multimodal approaches, incorporating visual and contextual cues, is discussed as a potential avenue for further improvement. Ethical considerations related to privacy and bias in speech recognition systems are also addressed, emphasizing the importance of responsible development and deployment. By synthesizing current research findings and anticipating future directions, this paper contributes to the evolving discourse on speech recognition technologies, providing a valuable resource for researchers, practitioners, and industry professionals in the field. Key Words: Real-time processing , Machine learning , Deep neural networks , Technology advancements , Contextual information , Large-scale datasets Transfer learning , End-to-end models , Real-time processing Edge computing , Multimodal approaches Ethical considerations , Privacy , Bias , Future prospects Research review.

Utilizing Deep Learning Techniques for the Classification of Spoken Languages in India

In Western countries, speech-recognition applications are accepted. In East Asia, it isn't as common. The complexity of the language might be one of the main reasons for this latency. Furthermore, multilingual nations such as India must be considered in order to achieve language recognition (words and phrases) utilizing speech signals. In the last decade, experts have been clamoring for more study on speech. In the initial part of the pre-processing step, a pitch and audio feature extraction technique were used, followed by a deep learning classification method, to properly identify the spoken language. Various feature extraction approaches will be discussed in this review, along with their advantages and disadvantages. Also discussed were the distinctions between various machine learning and deep learning approaches. Finally, it will point the way for future study in Indian spoken language recognition, as well as AI technology.

A Spiking LSTM Accelerator for Automatic Speech Recognition Application Based on FPGA

Long Short-Term Memory (LSTM) finds extensive application in sequential learning tasks, notably in speech recognition. However, existing accelerators tailored for traditional LSTM networks grapple with high power consumption, primarily due to the intensive matrix–vector multiplication operations inherent to LSTM networks. In contrast, the spiking LSTM network has been designed to avoid these multiplication operations by replacing multiplication and nonlinear functions with addition and comparison. In this paper, we present an FPGA-based accelerator specifically designed for spiking LSTM networks. Firstly, we employ a low-cost circuit in the LSTM gate to significantly reduce power consumption and hardware cost. Secondly, we propose a serial–parallel processing architecture along with hardware implementation to reduce inference latency. Thirdly, we quantize and efficiently deploy the synapses of the spiking LSTM network. The power consumption of the accelerator implemented on Artix-7 and Zynq-7000 is only about 1.1 W and 0.84 W, respectively, when performing the inference for speech recognition with the Free Spoken Digit Dataset (FSDD). Additionally, the energy consumed per inference is remarkably efficient, with values of 87 µJ and 66 µJ, respectively. In comparison with dedicated accelerators designed for traditional LSTM networks, our spiking LSTM accelerator achieves a remarkable reduction in power consumption, amounting to orders of magnitude.

Acoustic and Spectral Analysis of Adi Triphthongs

Adi, a zero-resource indigenous tribal language of Arunachal Pradesh, originated in Tibeto-Burman. According to the “UNESCO Atlas of the World's Languages in Danger of 2017” report, Adi is critically endangered. In this research, two new Adi triphthongs, /uai/ [uai] and /oai/ [ɔai], are proposed in the phoneme set of the Adi language. Adi has 51 phonemes, including 14 monophthongs (7 short and 7 long vowels), 19 diphthongs, two triphthongs, and 16 consonants. This research analysed Adi triphthongs’ spectral and acoustic properties, including intensity variation, long-term average spectrum (LTAS), and formant distribution. The mean of the first four formant frequencies of male and female Adi speakers is measured, which may be the identification of these two phones in the Adi speech recognition systems. Four different Adi words for each triphthong are recorded from 68 individual native Adi speakers (32 males and 36 females) in Arunachal Pradesh. This study may be helpful for phonetic modelling to build speech recognition applications for the Adi language.

Chinese dialect speech recognition: a comprehensive survey

As a multi-ethnic country with a large population, China is endowed with diverse dialects, which brings considerable challenges to speech recognition work. In fact, due to geographical location, population migration, and other factors, the research progress and practical application of Chinese dialect speech recognition are currently at different stages. Therefore, exploring the significant regional heterogeneities in specific recognition approaches and effects, dialect corpus, and other resources is of vital importance for Chinese speech recognition work. Based on this, we first start with the regional classification of dialects and analyze the pivotal acoustic characteristics of dialects, including specific vowels and tones patterns. Secondly, we comprehensively summarize the existing dialect phonetic corpus in China, which is of some assistance in exploring the general construction methods of dialect phonetic corpus. Moreover, we expound on the general process of dialect recognition. Several critical dialect recognition approaches are summarized and introduced in detail, especially the hybrid method of Artificial Neural Network (ANN) combined with the Hidden Markov Model(HMM), as well as the End-to-End (E2E). Thirdly, through the in-depth comparison of their principles, merits, disadvantages, and recognition performance for different dialects, the development trends and challenges in dialect recognition in the future are pointed out. Finally, some application examples of dialect speech recognition are collected and discussed.

CodeFed: Federated Speech Recognition for Low-Resource Code-Switching Detection

One common constraint in the practical application of speech recognition is Code Switching. The issue of code-switched languages is especially aggravated in the context of Indian languages – since most massively multilingual models are trained on corpora that are not representative of the diverse set of Indian languages. An associated constraint with such systems is the privacy-intrusive nature of the applications that aim to collate such representative data. To collectively mitigate both problems, this work presents CodeFed: A federated learning-based code-switching detection model that can be deployed to collaboratively be trained by leveraging private data from multiple users, without compromising their privacy. Using a representative low-resource Indic dataset, we demonstrate the superior performance of a collaboratively trained global model that is trained using federated learning on three low-resource Indic languages – Gujarati, Tamil and Telugu and draw a comparison of the model with respect to the most current work in the field. Finally, to evaluate the practical realizability of the proposed system, CodeFed also discusses the system overview of the label generation architecture which may accompany CodeFed’s possible real-time deployment.

Deep causal speech enhancement and recognition using efficient long-short term memory Recurrent Neural Network.

Long short-term memory (LSTM) has been effectively used to represent sequential data in recent years. However, LSTM still struggles with capturing the long-term temporal dependencies. In this paper, we propose an hourglass-shaped LSTM that is able to capture long-term temporal correlations by reducing the feature resolutions without data loss. We have used skip connections in non-adjacent layers to avoid gradient decay. In addition, an attention process is incorporated into skip connections to emphasize the essential spectral features and spectral regions. The proposed LSTM model is applied to speech enhancement and recognition applications. The proposed LSTM model uses no future information, resulting in a causal system suitable for real-time processing. The combined spectral feature sets are used to train the LSTM model for improved performance. Using the proposed model, the ideal ratio mask (IRM) is estimated as a training objective. The experimental evaluations using short-time objective intelligibility (STOI) and perceptual evaluation of speech quality (PESQ) have demonstrated that the proposed model with robust feature representation obtained higher speech intelligibility and perceptual quality. With the TIMIT, LibriSpeech, and VoiceBank datasets, the proposed model improved STOI by 16.21%, 16.41%, and 18.33% over noisy speech, whereas PESQ is improved by 31.1%, 32.9%, and 32%. In seen and unseen noisy situations, the proposed model outperformed existing deep neural networks (DNNs), including baseline LSTM, feedforward neural network (FDNN), convolutional neural network (CNN), and generative adversarial network (GAN). With the Kaldi toolkit for automated speech recognition (ASR), the proposed model significantly reduced the word error rates (WERs) and reached an average WER of 15.13% in noisy backgrounds.

(Invited) Physical Understanding of HfZrO2/Si FeFET Memory and Its AI Applications

Since the discovery of ferroelectricity in HfO2-based dielectric films in 2011 [1], FeFETs using HfO2-based ferroelectric films as gate dielectrics have attracted strong interest and have been intensively studied in the worldwide research community for various applications including memory, logic, and AI computing. In this paper, we present our recent activities on HfZrO2 (HZO)/Si FeFETs memory [2-6] and FeFET reservoir computing [7-9] for low-power memory and AI applications.HZO-based FeFETs are currently expected for a variety of applications such as embedded nonvolatile memory, cross-bar Commutation-in-Memory (CiM), and so on. However, the device operation has not been quantitatively understood yet. One of the fundamental questions in FeFETs is a mismatch between areal surface carrier concentration (Ns) in MOSFETs, ~1013 cm-2 at maximum, and typical polarization (P) values, ~10-20 mC/cm2. These polarization values correspond to Ns higher than 1014 cm-2 and induce an electric field across an interfacial layer (IL) of 30 MV/cm or higher, which could cause oxide breakdown. On the other hand, when a large amount of traps is included in an HZO film or an HZO/IL interface, sufficient P values can be obtained without too high Ns and the electric field [2, 4-6]. Thus, it is very important to experimentally characterize P, Ns, and trapped charge density, which were evaluated by P-V, quasi-static split C-V, and Hall measurements, respectively, [2, 3, 6] for 10-nm-thick HZO/Si n- and p-FeFETs with 0.7-nm-thick SiO2 IL. As a result, it has been found that a high trap density of around 1014 cm-2 is observed in n-FeFETs, while there is almost no hole trapping in p-FeFETs. This high electron trap density leads to the high electric field across the HZO film and resulting high P values in n-FeFETs, which is in contrast to p-FETs. Thus, the larger memory window in n-FeFETs than that in p-FeFET is attributed to the difference in P values with applying the positive and negative gate voltages.On the other hand, we have proposed and demonstrated reservoir computing, which is one of AI calculations using a recurrent neural network, by using HZO/Si FeFETs [7-9]. Reservoir computing has recently been attracting strong attention as a method to realize real-time learning of time-series data at the edge. Furthermore, when this reservoir is represented by hardware that has input-history-dependent and nonlinear dynamics (physical reservoir), we can expect significant improvement in the energy efficiency of AI calculations. In our FeFET reservoir, the time response of a drain current of Si FeFETs was utilized as the virtual nodes. We have experimentally demonstrated high classification accuracy in speech recognition by using this HZO/Si FeFET reservoir computing with parallel data processing [9]. The spoken digit recognition task, which is an example of speech recognition applications, was performed on a basis of multiple frequency channel (cochleagram) data, which were input to multiple FeFETs. The final inference was conducted by a majority vote of the multiple FeFET reservoirs. We introduced several approaches in the proposed FeFET reservoir computing system, such as optimization of the number of virtual nodes per time step, analog and digital input waveforms, the utilization of other currents than the drain current, and the combination of different frequency channels, in order to improve the classification accuracy. The choice of a combination of frequency channels was effective in improving recognition accuracy. Also, combining the time responses of the drain, source, and substrate currents was also important. These various methods additively improved the recognition accuracy. As a result, we have experimentally demonstrated a classification accuracy of 95.9 % in the task to classify the audio waveforms of ‘0’ to ‘9’ spoken digits.This work was supported by JST CREST (JPMJCR20C3) and JSPS KAKENHI (21H01359).

Environment-Aware Knowledge Distillation for Improved Resource-Constrained Edge Speech Recognition

Recent advances in self-supervised learning have allowed automatic speech recognition (ASR) systems to achieve state-of-the-art (SOTA) word error rates (WER) while requiring only a fraction of the labeled data needed by its predecessors. Notwithstanding, while such models achieve SOTA results in matched train/test scenarios, their performance degrades substantially when tested in unseen conditions. To overcome this problem, strategies such as data augmentation and/or domain adaptation have been explored. Available models, however, are still too large to be considered for edge speech applications on resource-constrained devices; thus, model compression tools, such as knowledge distillation, are needed. In this paper, we propose three innovations on top of the existing DistilHuBERT distillation recipe: optimize the prediction heads, employ a targeted data augmentation method for different environmental scenarios, and employ a real-time environment estimator to choose between compressed models for inference. Experiments with the LibriSpeech dataset, corrupted with varying noise types and reverberation levels, show the proposed method outperforming several benchmark methods, both original and compressed, by as much as 48.4% and 89.2% in the word error reduction rate in extremely noisy and reverberant conditions, respectively, while reducing by 50% the number of parameters. Thus, the proposed method is well suited for resource-constrained edge speech recognition applications.

High-bandwidth self-powered vibration sensors based on triboelectric particle-surface interactions

Vibrating triboelectric nanogenerators have the potential as self-powered vibration sensors if they can be designed to operate over a wide frequency band. Herein, we present a high-bandwidth self-powered triboelectric particle-surface vibration sensor. The sensor consists of a flat round capsule, filling with dielectric particles. The inner surfaces of the capsule are made of triboelectric materials backed by electrodes. When attached to a vibrating structure, the impact of the particles on the inner surfaces of the capsule causes them to be oppositely charged, and the dynamic motion of the charged particles in turn generates an alternating voltage across the two electrodes. A systematic investigation of the effects of particle properties, including material type, diameter, and packing ratio on the sensor’s output performance revealed that, among several designs, the combination of nylon particles and PTFE triboelectric layers yielded the highest output performance. The sensor’s flat frequency response has been found to span between 500 Hz and 7000 Hz, which is significantly higher than the recently reported values for triboelectric vibration sensors. Furthermore, the novel vibration sensor offers a frequency resolution of 0.19% at frequencies up to 8 kHz. The potentials of this new type of vibration sensor have been demonstrated for speech recognition applications and to detect structural vibrations due to drop-weight impact at different distances and heights.

A Review on Speech Recognition for Under-Resourced Languages

Fundamental speech recognition technologies for high-resourced languages are currently successful to build high-quality applications with the use of deep learning models. However, the problem of “borrowing” these speech recognition technologies for under-resourced languages like Vietnamese still has challenges. This study reviews fundamental studies on speech recognition in general as well as speech recognition in Vietnamese, an under-resourced language in particular. Then, it specifies the urgent issues that need current research attention to build Vietnamese speech recognition applications in practice, especially the need to build an open large sentence-labeled speech corpus and open platform for related research, which mostly benefits small individuals/organizations who do not have enough resources.

Ultrasensitive Acoustic Detection Using an Enlarged Fabry-Perot Cavity with a Graphene Diaphragm.

For exerting high sensitivity of ultrathin graphene to detection deformation, an enlarged backing air cavity (EBC) structure is developed to further enhance the mechanical sensitivity (SM) of a graphene-based Fabry-Perot (F-P) acoustic sensor. COMSOL acoustic field simulation on the air cavity size-dependent SM confirms the optimal length and radius of the EBC of 0.2 and 1.5 mm, respectively, with the maximum simulation SM of 26.16 nm/Pa@1 kHz. Acoustic experiments further demonstrate that the frequency response of the fabricated graphene-based F-P acoustic sensor after the use of the EBC is enhanced by 5.73-79.33 times in the range of 0.5-18 kHz, compared with the conventional one without the EBC. Especially the maximum SM is up to 187.32 nm/Pa@16 kHz, which is at least 17% higher than the SM values ranging from 1.1 to 160 nm/Pa in previously reported F-P acoustic sensors using various diaphragm materials. More acoustic characteristics are examined to highlight various merits of the EBC structure, including a signal-to-noise ratio (SNR) of 60-75 dB@0.5-18 kHz, a time stability of less than ±1.3% for 90 min, a detection resolution of 0.01 Hz, and a high-fidelity speech detection with a cross-correlation coefficient of greater than 0.9, thereby revealing its high-performance weak acoustic sensing and speech recognition applications.

Research on the application of machine learning models in speech recognition in noisy environments

In a world increasingly reliant on verbal communication with machines, this paper undertakes a rigorous exploration of the application of machine learning models in speech recognition within noisy environments. An important focus of this research lies in demonstrating the viability and robustness of these models when challenged by real-world noise interference. In particular, the utilization of deep learning models such as Convolutional Neural Networks (CNNs) and Long Short-Term Memory Networks (LSTMs) is thoroughly examined. Using concrete data and case studies, this paper uncovers a promising trajectory in the development and implementation of noise-robust speech recognition systems. In so doing, it lays a foundation for future research to enhance the practicality and effectiveness of machine learning models in this area. The ultimate goal is to ensure more reliable and accurate speech recognition that can withstand various noise conditions, thus expanding the potential for speech recognition applications in diverse and adverse environments.

Application of multiple difference feature network and speech recognition in dance training system

Application of multiple difference feature network and speech recognition in dance training system

Assessment of the Perception and Worries of Saudi Healthcare Providers About the Application of Artificial Intelligence in Saudi Health Facilities.

Objective This study is aimed at assessing the perception and worries of Saudi healthcare providers about the application of artificial intelligence (AI) in Saudi healthcare facilities. Methods The study adopted a cross-sectional study involving 1026 Saudi healthcare providers between January 2023 and April 2023. The target population was healthcare providers across Saudi health facilities. Online questionnaires were administered through social media platforms. Data were analyzed using SPSS Statistics, version 26.0 (IBM Corp., Armonk, NY) to obtain important insights. Results The results of this study indicated that more than half (55.2%) of the respondents had good knowledge of AI, with (48.1%) of them being familiar with the application of AI in their specialty. A good proportion of the participants (57.9%) knew at least one term about the difference between machine learning and deep learning. More than half (69.9%) of the participants indicated that they had at one point in time used speech recognition or transcription application in their work. A large section (73.3%) of healthcare providers believed that AI would replace them at their job. A vast majority (84.9%) of the participants agreed that collaboration between medical schools with engineering and computer science faculties could be a game changer to provide a road for incorporating AI into medical curricula. The mean perception of AI in this study was 37.6 (SD=8.41; range 0-241). Age, level of health, health profession, and working experience all significantly impacted the positive perception score (p=0.021; p=0.031; p=0.041; p=0.026). However, there was no significant association between gender, nationality, and Saudi regions with a mean positive perception score. Conclusion There was a positive perception of AI among Saudi healthcare providers. Even though a substantial majority of Saudi healthcare providers were worried that AI would replace their jobs, the study revealed that AI serves as a crucial practitioner's tool rather than a physician's replacement.