Speech Analysis Research Articles

Analisis Ilokusi dan Implikaturnya pada Teks Iklan, Slogan, dan Poster dalam Materi Bahasa Indonesia Kelas VIII Kurikulum Merdeka

This research aims to analyze illocutionary and implicatures in advertisments, slogans, and posters in the Indonesian language material for class VIII Merdeka Curriculum. This research uses a descriptive qualitative and pragmatic approach. The authors obtained data by listening and reading related references and recording information that could be used a data. For this reason, the techniques used in this research are listening, reading, and note-taking. The data studied in this research are advertisments, slogans, and posters in the Indonesian language material for class VIII Merdeka Curriculum. The data analysis method used in this article method and pragmatic heuristic method. Researchers analyzed the text in depth in terms of characteristics and truth testing. As a result of data analysis, only three types of five types of illocutionary speech acts were found in the 12 samples on advertisment images, slogans, and posters, including (1) assertive speech acts of stating and providing information; (2) inviting directive speech acts; and (3) prohibit declarative speech acts. Through this research, it is hoped that we will be able to obtain an overview of illocutionary speech acts and their implicatures found in advertisments, slogans, and posters and practically this research can be used as a basis for future researchers, especially regarding the analysis of illocutionary speech acts and their implicatures.

Automated analysis of speech as a marker of sub-clinical psychotic experiences.

Automated speech analysis techniques, when combined with artificial intelligence and machine learning, show potential in capturing and predicting a wide range of psychosis symptoms, garnering attention from researchers. These techniques hold promise in predicting the transition to clinical psychosis from at-risk states, as well as relapse or treatment response in individuals with clinical-level psychosis. However, challenges in scientific validation hinder the translation of these techniques into practical applications. Although sub-clinical research could aid to tackle most of these challenges, there have been only few studies conducted in speech and psychosis research in non-clinical populations. This work aims to facilitate this work by summarizing automated speech analytical concepts and the intersection of this field with psychosis research. We review psychosis continuum and sub-clinical psychotic experiences, and the benefits of researching them. Then, we discuss the connection between speech and psychotic symptoms. Thirdly, we overview current and state-of-the art approaches to the automated analysis of speech both in terms of language use (text-based analysis) and vocal features (audio-based analysis). Then, we review techniques applied in subclinical population and findings in these samples. Finally, we discuss research challenges in the field, recommend future research endeavors and outline how research in subclinical populations can tackle the listed challenges.

Exploring age-related changes in acoustic voice analysis parameters: insights from a study on older people

Exploring age-related changes in acoustic voice analysis parameters: insights from a study on older people

Abstract TP78: Characteristics of Verbal and Non-Verbal Cognitive Function in Post-Stroke Aphasia Based on Language-Cognition Neuropsychological Processing Model

Introduction: Post-stroke aphasia (PSA) patients have different levels of phonetic, lexical, semantic, syntactic, and other linguistic domain impairment, and accompanied by non-verbal cognitive impairment. However, the association between verbal and non-verbal cognitive functions is not fully understood.In this study, we used the Chinese Aphasia Language Battery (CALB) and Loewenstein Occupational Therapy Cognitive Assessment (LOTCA) to fully evaluate the characteristics of verbal and non-verbal cognitive function impairment in PSA patients. Methods: A total of 105 stroke patients were recruited in this study. Language and non-linguistic cognitive impairments were evaluated by CALB and LOTCA, respectively. Patients in the PSA group were subdivided into the fluent aphasia group (FAG) and the non-FAG group. Potential confounders were adjusted in the analysis of covariance. Partial correlation analysis between the subscores of CALB and LOTCA was also performed. Results: This study showed that PSA patients had worse speech analysis ability, vocabulary processing ability, semantic processing ability, syntactic structure conversion ability, and non-verbal cognitive performance(P<0.001). Moreover, the scores of phonetic decoding, phonetic vocabulary production, noun category classification, total LOTCA, orientation, visual perception, motor use, and thinking operation were lower in the non-FAG compared with the FAG (p<0.05). The total score of LOTCA was positively correlated with the aphasia quotient (r=0.637, P<0.001) and scores of speech analysis ability, vocabulary processing ability, semantic processing ability, and syntactic structure conversion ability (r>0; P<0.0005). Conclusions: Based on the language-cognition neuropsychological processing model, this study adopted a comprehensive verbal and non-verbal cognitive assessment set of tests, and found that PSA patients have extensive phonological, lexical, semantic, syntactic, and grammatical processing disorders. Moreover, PSA patients often have more extensive and serious non-verbal cognitive impairments, and the severity of aphasia is significantly positively correlated with the degree of nonverbal cognitive impairment.

Prominence Analysis of President Jokowi's Speeches on the Palestine–Israel Conflict: Acoustic Phonetic Study

The aim of this research is to represent the prominence contained in President Jokowi's statements regarding the Palestinian and Israeli conflicts, as well as to find the values of frequency, duration, and intensity as characteristics of the sound. This research is quantitave research with an instrumental approach. The research data is an audiovisual form, and it is analyzed using the Praat application. The result shows that the prominence of President Jokowi's statement is found in the words 'indonesia' and 'war', namely in the rightmost syllables: -sia and -rang. When communicating, speakers tend to pay more attention to words that they think are important. It has also been proven that the factors that influence the formation of prominence are pitch, loudness, and duration, with values of F0, intensity, and duration for stressed syllables is higher than unstressed syllables. Apart from that, prominence can also occur when speakers want to highlight words that are considered important.

Ideologies on Display: A Critical Analysis of First-Term Gubernatorial Inaugural Speeches in Nigeria

The interest of scholars in political discourse has always been drawn to the dynamics of inaugural speeches and the implicit/explicit portrayal of ideologies. Extant studies have explored discourse patterns, thematic orientations, power relations, rhetorical devices, ideologies, and other issues in inaugural speeches. This paper offers a holistic investigation of the recurring ideologies in first-term gubernatorial inaugural speeches across Nigeria. The data for the study are drawn from 25 inaugural speeches made between 2014 and 2015 across the 36 states in Nigeria. The source of the data is Inaugural Speeches: President and Governors of Nigeria (2014-2017). Through the theoretical lens of van Dijk’s approach to critical discourse analysis, four ideologies were identified in the speeches. These manifest ideologies in the speeches are idealistic ideology, theistic ideology, messianic ideology, and democratic ideology. They are found to drive the establishment of the authority of the governors through the process of sensitizing the citizens about the governor’s interpretation of the election, the presentation of novel principles and perspectives that the new administration would imbibe, and how they would inform governance for the tenure. The reoccurring strategies actualized to project the ideologies are consensus, polarization, lexicalization, presupposition, and metaphors.

Comparative analysis of voice denoising using machine learning and traditional denoising

Noise often affects the content of an audio signal, and noise reduction techniques can help retrieve the original speech content. In recent years, AI-based noise reduction has witnessed rapid development. This article provides a brief introduction to the background and principles of several AI-based noise reduction methods. One of the mentioned methods is an end-to-end time-domain deep learning speech division algorithm, which utilizes a multi-layer CNN network framework. Due to the need for deep network architectures to extract features, it involves a higher computational load. Traditional noise reduction algorithms, on the other hand, are based on researchers' understanding of noise patterns and modeling. Traditional methods may not perform well on non-stationary noise, but they are relatively simple in terms of algorithmic implementation. Through a comparison from various perspectives, AI-based noise reduction demonstrates superior performance in known environments compared to traditional methods. However, in unknown environments, AI-based noise reduction may encounter performance anomalies. Combining AI-based and traditional noise reduction techniques can provide better stability and higher performance in certain scenarios.

A NEW APPROACH FOR SPEECH EMOTION RECOGNITION USING SINGLE LAYERED CONVOLUTIONAL NEURAL NETWORK

Creating a computational device to identify human emotions via voice analysis represents a notable achievement in the sector of human-computer interaction, especially within the healthcare domain. We propose a new light-weight model for addressing challenges of emotions recognition. The model works based on CNN with change of kernel processing. The proposed model performs a direct matching to recognize speech emotions of different eight categories using a statistical model named Analysis of Variance (ANOVA) as kernel for features extraction and Cosine Similarity Measurement (CSM) as activation function for CNN model. This proposed model contains eight-folded single-layered intermediate neurons, and each neuron can segregate speech emotion pattern using CSM from the voice convergence matrix to explore a part of the solution from the whole solution. Experiment results demonstrates that the proposed model outperforms compared with multiple layered existing CNN methods in identifying the emotional state of a speaker.

Critical swallowing functions contributing to dysphagia in patients with recurrent laryngeal nerve paralysis after esophagectomy.

Recurrent laryngeal nerve paralysis (RLNP) after esophagectomy can cause aspiration because of incomplete glottis closure, leading to pneumonia. However, patients with RLNP often have preserved swallowing function. This study investigated factors that determine swallowing function in patients with RLNP. Patients with esophageal cancer who underwent esophagectomy and cervical esophagogastric anastomosis were enrolled between 2017 and 2020. Videofluoroscopic examination of swallowing study (VFSS) and acoustic voice analysis were performed on patients with suspected dysphagia including RLNP. Dysphagia in VFSS was defined as score ≥ 3 of the 8-point penetration-aspiration scale VFSS and acoustic analysis results related to dysphagia were compared between patients with and without RLNP. Among 312 patients who underwent esophagectomy, 74 developed RLNP. The incidence of late-onset pneumonia was significantly higher in the RLNP group than in the non-RLNP (18.9 vs. 8.0%, P = .008). Detailed swallowing function was assessed by VFSS in 84 patients, and patients with RLNP and dysphagia showed significantly shorter maximum diagonal hyoid bone elevation (10.62 vs. 16.75mm; P = .003), which was a specific finding not seen in patients without RLNP. For acoustic voice analysis, the degree of hoarseness was not closely related to dysphagia. The length of oral intake rehabilitation for patients with and without RLNP was comparable if they did not present with dysphagia (8.5 vs. 9.0days). Impaired hyoid bone elevation is a specific dysphagia factor in patients with RLNP, suggesting compensatory epiglottis inversion by hyoid bone elevation is important for incomplete glottis closure caused by RLNP.

A NEW APPROACH FOR SPEECH EMOTION RECOGNITION USING SINGLE LAYERED CONVOLUTIONAL NEURAL NETWORK

Creating a computational device to identify human emotions via voice analysis represents a notable achievement in the sector of human-computer interaction, especially within the healthcare domain. We propose a new light-weight model for addressing challenges of emotions recognition. The model works based on CNN with change of kernel processing. The proposed model performs a direct matching to recognize speech emotions of different eight categories using a statistical model named Analysis of Variance (ANOVA) as kernel for features extraction and Cosine Similarity Measurement (CSM) as activation function for CNN model. This proposed model contains eight-folded single-layered intermediate neurons, and each neuron can segregate speech emotion pattern using CSM from the voice convergence matrix to explore a part of the solution from the whole solution. Experiment results demonstrates that the proposed model outperforms compared with multiple layered existing CNN methods in identifying the emotional state of a speaker.

Speech Analysis Classification and the Form of Commissive Speech Act Used in the Song "Beautiful Day" by Jermaine Edwards

A speech act is an action performed when saying something. Without taking into account both the hearers and the speakers, speech acts cannot be properly comprehended. Speech acts are intended for real people, whose capacity for recognition places restrictions on the expressiveness of the lyrics. This research aimed to analyze the speech act classifications and the forms of commissive speech act used in Jermaine Edwards' song "Beautiful Day". This research used a qualitative method. The object of the research was Jermaine Edwards' song "Beautiful Day." The data were collected by using the observational method, followed by listening and note-taking techniques. Then the data were analyzed by using a descriptive approach to describe the types and forms of descriptive speech acts found in the song lyrics. The researcher used the theory of Searle (1969) to analyze the speech act classifications and the forms of commissive speech acts. After analyzing the data, it was found five types of speech acts: representative, declarative, directive, expressive, and commissive. The forms of commissive speech act in the song "Beautiful Day" by Jermaine Edwards were promise, warn, and threat.

Author identification of literary works based on text analysis and deep learning

With the development of science, speech, picture, and other analysis, problems have been gradually better solved, but the study of Chinese text has been a complex problem to overcome. Chinese text analysis requires not only statistics but also semantic comprehension analysis. Different text types need other language style feature modeling to obtain good recognition results. In this study, we use the deep learning method to construct an automatic text feature extraction model and classify it with the author as a classification label. This study presents a literature author recognition model based on deep learning, which is mainly divided into three phases: text preprocessing, feature extraction, and classification. Each part consists of several small modules or steps. First, we input the corpus to Word2Vec to generate the new word vector. Then, the improved text feature extractor based on CNN and Attention extracts the text features and uses them as the input of the CNN convolution layer. After convolution, the text is combined with bits to get Window Feature Sequence. It is the text feature vector. Next, based on LSTM and Softmax classification output, Window Feature Sequence is used as the input of LSTM to obtain two one-dimensional vectors spliced by concatenate layer. Finally, the result is classified through the fully connected layer, Batch Normalization layer, and Softmax. The performance of the proposed model in recognizing authors of Chinese literature was evaluated using two datasets. In the research process, the data we collected included works of different forms, such as prose and fiction. The research results show that the proposed model can effectively identify author identity. The classification accuracy of our proposed algorithm is significantly better than that of the benchmark model.

Research On Emotion Management Based on Speech Analysis for Nursing Homes

Speech emotion recognition is an important research area in artificial intelligence aimed at identifying the emotional states of speakers. This paper provides an overview of the current state and key algorithms of speech emotion recognition, focusing particularly on the emotional well-being of elderly residents in nursing homes. Firstly, the paper introduces the background and application areas of speech emotion recognition, emphasizing its significance in human-computer interaction, psychological health monitoring, and emotional intelligent systems. The paper extensively discusses methods of extracting emotional features, it adopts the K-Nearest Neighbors (KNN) classification algorithm, which serves as a simple yet effective classification method with potential in emotional speech recognition. This study aims to explore the design, implementation, and future directions of an emotion recognition system based on KNN. This algorithm exhibits a high fitting performance for emotional speech data, presenting advantages such as ease of implementation and alignment with the distribution characteristics of emotional speech data. Through multiple comparative experiments, the optimal K value for recognition has been determined, achieving a straightforward and convenient emotional recognition simulation. Finally, the paper summarizes the current state of speech emotion recognition, emphasizing its potential and significance in practical applications.

Ideological Disagreement and the Rejection of Laws by Italian Heads of State

AbstractHeads of state in parliamentary democracies are usually perceived as neutral and non‐partisan political actors. However, they can make autonomous political choices, sometimes oriented by their own partisan preferences. Focusing on the Italian case, we analyze the propensity of the president of the Republic to publicly express political disagreement by rejecting the promulgation of laws. We contribute to the literature by creating and exploiting a new dataset, based on a content analysis of presidents' investiture speeches and New Year's Eve messages which maps the ideological position of Italian heads of state and their distance from the cabinet over more than 75 years (1946–2022). We show that such ideological distance increases the number of laws rejected by the president. This effect holds when controlling for the size and heterogeneity of the government coalition, as well as the state of the economy and the occurrence of crises.

Machine Learning-Based Classification of Parkinson's Disease Patients Using Speech Biomarkers.

Parkinson's disease (PD) is the most prevalent neurodegenerative movement disorder and a growing health concern in demographically aging societies. The prevalence of PD among individuals over the age of 60 and 80 years has been reported to range between 1% and 4%. A timely diagnosis of PD is desirable, even though it poses challenges to medical systems. This study aimed to classify PD and healthy controls based on the analysis of voice records at different frequencies using machine learning (ML) algorithms. The voices of 252 individuals aged 33 to 87 years were recorded. Based on the voice record data, ML algorithms can distinguish PD patients and healthy controls. One binary decision variable was associated with 756 instances and 754 attributes. Voice records data were analyzed through supervised ML algorithms and pipelines. A 10-fold cross-validation method was used to validate models. In the classification of PD patients, ML models were performed with 84.21 accuracy, 93 precision, 89 Sensitivity, 89 F1-scores, and 87 AUC. The pipeline performance improved to accuracy: 85.09, precision: 92, Sensitivity:91, F1-score: 89, and AUC: 90. The Pipeline methods improved the performance of classifying PD from voice record. Our study demonstrated that ML classifiers and pipelines can classify PD patients based on speech biomarkers. It was found that pipelines were more effective at selecting the most relevant features from high-dimensional data and at accurately classifying PD patients and healthy controls. This approach can therefore be used for early diagnosis of initial forms of PD.

Human Digital Twin in Industry 5.0: A Holistic Approach to Worker Safety and Well-Being through Advanced AI and Emotional Analytics.

This research introduces a conceptual framework designed to enhance worker safety and well-being in industrial environments, such as oil and gas construction plants, by leveraging Human Digital Twin (HDT) cutting-edge technologies and advanced artificial intelligence (AI) techniques. At its core, this study is in the developmental phase, aiming to create an integrated system that could enable real-time monitoring and analysis of the physical, mental, and emotional states of workers. It provides valuable insights into the impact of Digital Twins (DT) technology and its role in Industry 5.0. With the development of a chatbot trained as an empathic evaluator that analyses emotions expressed in written conversations using natural language processing (NLP); video logs capable of extracting emotions through facial expressions and speech analysis; and personality tests, this research intends to obtain a deeper understanding of workers' psychological characteristics and stress levels. This innovative approach might enable the identification of stress, anxiety, or other emotional factors that may affect worker safety. Whilst this study does not encompass a case study or an application in a real-world setting, it lays the groundwork for the future implementation of these technologies. The insights derived from this research are intended to inform the development of practical applications aimed at creating safer work environments.

A comparative study of short-term phonatory outcomes in primary cases of early glottic cancer treated with radiotherapy versus laser surgery

BackgroundEarly-stage glottic carcinomas can be treated with either laser surgery or radiotherapy. Both approaches have demonstrated similar cure rates. However, the question of which modality is superior in terms of voice outcomes remains a topic of debate.MethodsIn our study, we conducted a comparison of short-term phonatory outcomes in patients with primary early glottic cancer who underwent treatment using both modalities. We assessed these outcomes using videostroboscopy and voice analysis software at three time points: immediately following treatment, at 1 month, and at 3 months post-treatment.ResultsVoice analysis revealed that laser surgery had a more favorable immediate effect compared to radiotherapy in parameters such as jitter, shimmer, harmonics-to-noise ratio, and fundamental frequency. However, at the 3-month post-treatment mark, both treatment modalities demonstrated similar effects. Additionally, both modalities had comparable effects on maximum phonation time. Videostroboscopy observations showed that mucosal wave forms were more prominent immediately after laser surgery and gradually improved following radiotherapy. Furthermore, patients with incomplete glottic closure experienced recovery after both treatment modalities.ConclusionBased on the results, there are better outcomes immediately following LS compared to RT. However, after a 3-month period, the outcomes of both treatment modalities become comparable. The treating physicians must consider various factors such as complications, patient-specific considerations, treatment costs, and duration to make informed decisions. A personalized approach considering the individual patient’s circumstances is crucial in achieving optimal results in the management of T1 laryngeal cancer.

The Study of Spontaneity and Preparedness of Oral Speech in Forensic Linguistics: To the Formulation of the Problem

The article considers different approaches to the study of spontaneity and readiness of speech – in linguistics and forensic linguistics.In linguistics, there are two directions of studying the spontaneity and preparedness of sounding speech: the study of the speech flow as a whole and the analysis of speech by language levels. In the first case, the main attention is paid to the ontogenesis of speech thinking, the formation of a person’s communicative competence, as well as the mechanisms of speech generation. In the level-by-level analysis of oral speech, the study of the phonetic level of the language occupies a leading place, while research at other levels of the language is presented unsystematically. In the practice of forensic examination, gradations of sounding speech are determined according to the criterion «prepared speech – unprepared speech». The parameters that are taken into account in the study of sounding speech for the identification of signs of preparedness and spontaneity are established: type (form) of speech; genre in which the speech is performed; age of the speaker; the speaker’s level of education, his professional activity, social status; parameters of the communicative situation: the addresser and the addressee, the relationship between the speaker and the listener, the tone of communication, the purpose, means of communication, the place of communication, etc.; the nature of communication.The author of the present article attempts to develop possible directions for the study of sounding speech in order to determine its spontaneity and preparedness.

A Comparative Rhetorical Analysis of Trump and Biden's Climate Change Speeches: Framing Strategies in Politics

This article analyzes the speeches of two U.S. politicians—President Donald Trump and President Joe Biden—to present how they make their arguments about climate change using various frames. While frames are rhetorical acts, they are also a form of persuasion. In particular, the author demonstrates how Trump foregrounded negative frames with fear-inducing elements. He presented job losses and economic harm as consequences of joining the Paris Climate Accord, putting him on the defensive. In contrast, Biden utilized positive frames to strengthen his arguments and aligned more closely with the environmental justice framework. Inspired by the rhetoric of the framing strategies employed by these two speakers, the study suggests that technical communicators should focus on using language that constructs new frames to enhance the success of their argumentations.

Multilingual Speaker Recognition using Mel-frequency Cepstral Coefficients and Gaussian Mixture Model

Introduction: People can recognize a speaker with the help of their voice via mobile or digital devices. Method: To obtain this congenital human being ability, authentication techniques based on speaker biometrics like automated speaker recognition (ASR) have been proposed. An ASR identifies speakers by speech signals analysis and salient feature extraction from their voices. Result: This will become an important part of recent research in the voice biometrics field. This paper proposes multilingual speaker recognition with the help of MFCC as feature extraction and GMM as classification techniques using various available datasets such as TIMIT, librespeech, etc. Conclusion: The results achieved from the given datasets enhance the recognition rate of 70.98% with MFCC.