Speaker Characteristics Research Articles

Body Perception From Connected Speech: Speaker Height Discrimination from Natural Sentences and Sine-Wave Replicas with and without Pitch.

In addition to language, the human voice carries information about the physical characteristics of speakers, including their body size (height and weight). The fundamental speaking frequency, perceived as voice pitch, and the formant frequencies, or resonators of the vocal tract, are the acoustic speech parameters that have been most intensely studied for perceiving a speaker's body size. In this study, we created sine-wave (SW) replicas of connected speech (sentences) uttered by 20 male and 20 female speakers, consisting of three time-varying sinusoidal waves matching the frequency pattern of the first three formants of each sentence. These stimuli only provide information about the formant frequencies of a speech signal. We also created a new experimental condition by adding a sinusoidal replica of the voice pitch of each sentence. Results obtained from a binary discrimination task revealed that (a) our SW replicas provided sufficient useful information to accurately judge the speakers' body height at an above chance level; (b) adding the sinusoidal replica about the voice pitch did not significantly increase accuracy; and (c) stimuli from female speakers were more informative for body height detection and allowed higher perceptual accuracy, due to a stronger correlation between formant frequencies and actual body height than stimuli from male speakers.

End-to-End Speaker Profiling Using 1D CNN Architectures and Filter Bank Initialization

The automatic estimation of speaker characteristics, such as height, age, and gender, has various applications in forensics, surveillance, customer service, and many human-robot interaction applications. These applications are often required to produce a response promptly. This work proposes a novel approach to speaker profiling by combining filter bank initializations, such as continuous wavelets and gammatone filter banks, with one-dimensional (1D) convolutional neural networks (CNN) and residual blocks. The proposed end-to-end model goes from the raw waveform to an estimated height, age, and gender of the speaker by learning speaker representation directly from the audio signal without relying on handcrafted and pre-computed acoustic features. The conducted experiments on the TIMIT dataset show that the proposed approach outperforms many previous studies on speaker profiling with a mean absolute error (MAE) of 5.18 and 4.91 cm in height estimation and MAE of 5.36 and 6.07 years in age estimation for males and females, respectively, and achieving an accuracy of 99.98% in gender prediction.

Self-supervised contrastive speaker verification with nearest neighbor positive instances

Self-supervised contrastive learning (SSCL) has achieved a great success in speaker verification (SV). All recent works treat within-utterance speaker embeddings (SE) to be positive instances, encouraging them to be as close as possible. However, positive instances from the same utterance have similar channel and related semantic information, which are difficult to distinguish from the speaker features. Moreover, these positive instances can only provide limited variances in a certain speaker. To tackle the above problems, we propose to use nearest neighbor (NN) positive instances for training, which are selected from a dynamic queue. The NN positive instances can provide different channel and semantic information, increasing the variances in a certain speaker. Our proposed method are validated through comprehensive experiments on VoxCeleb and CNCeleb1 datasets, demonstrating its effectiveness in improving both SSCL and fine-tuning results. Additionally, our SSCL model outperforms supervised training model in cross-dataset testing due to the use of massive unlabeled data.

Psycholinguistic Approach: In the Case of Foreign Language Teaching

The article provides information about the fact that learning a foreign language with a set of methods of various disciplines is an urgent task of scientific knowledge. The study of the ability to learn foreign languages is carried out using psycholinguistic methods based on contractual forms of communication, taking into account the individual characteristics of speakers, as well as the level of their language culture, semantics and elements of speech.

LEXIS OF PAKISTANI ENGLISH: A STUDY OF LEXICAL BORROWING IN PAKISTANI ANGLOPHONE LITERATURE

Pakistani English, an institutionalized variation of English, developed as a result of language contact between English and the Pakistani language. One of the most notable features of this variant is the significant creativity in the lexis. In this study, based on Shah's, A Season for Martyrs (2014), and Mueenuddin's, In Other Rooms, Other Wonders (2009) the researchers explored the language's vocabulary through the lexical borrowings of single words and their assimilation in Pakistani English. They used qualitative content analysis method to analyze lexical borrowings of single items and their level of integration into Pakistani English, on a three-fold grade-ability scale, in the selected works. The study’s findings posit that Pakistani English frequently borrows lexical items from Pakistani languages either to fill the lexical gaps or to transmit various shades of meaning, which reinforces the notion that this variety has its own distinctive lexicon. In addition, the study endorsed the notion that Pakistani English should be considered an independent variety since it represents the linguistic and cultural characteristics of English speakers in Pakistan and is widely used in Pakistani Anglophone literature. Keywords. Pakistani English; Lexical Borrowing; Contextual Areas; Lexical Integration.

Robust and language-independent acoustic features in Parkinson's disease.

The analysis of vocal samples from patients with Parkinson's disease (PDP) can be relevant in supporting early diagnosis and disease monitoring. Intriguingly, speech analysis embeds several complexities influenced by speaker characteristics (e.g., gender and language) and recording conditions (e.g., professional microphones or smartphones, supervised, or non-supervised data collection). Moreover, the set of vocal tasks performed, such as sustained phonation, reading text, or monologue, strongly affects the speech dimension investigated, the feature extracted, and, as a consequence, the performance of the overall algorithm. We employed six datasets, including a cohort of 176 Healthy Control (HC) participants and 178 PDP from different nationalities (i.e., Italian, Spanish, Czech), recorded in variable scenarios through various devices (i.e., professional microphones and smartphones), and performing several speech exercises (i.e., vowel phonation, sentence repetition). Aiming to identify the effectiveness of different vocal tasks and the trustworthiness of features independent of external co-factors such as language, gender, and data collection modality, we performed several intra- and inter-corpora statistical analyses. In addition, we compared the performance of different feature selection and classification models to evaluate the most robust and performing pipeline. According to our results, the combined use of sustained phonation and sentence repetition should be preferred over a single exercise. As for the set of features, the Mel Frequency Cepstral Coefficients demonstrated to be among the most effective parameters in discriminating between HC and PDP, also in the presence of heterogeneous languages and acquisition techniques. Even though preliminary, the results of this work can be exploited to define a speech protocol that can effectively capture vocal alterations while minimizing the effort required to the patient. Moreover, the statistical analysis identified a set of features minimally dependent on gender, language, and recording modalities. This discloses the feasibility of extensive cross-corpora tests to develop robust and reliable tools for disease monitoring and staging and PDP follow-up.

ADAPT-TTS: HIGH-QUALITY ZERO-SHOT MULTI-SPEAKER TEXT-TO-SPEECH ADAPTIVE-BASED FOR VIETNAMESE

Current adaptive-based speech synthesis techniques are based on two main streams: 1. Fine-tuning the model using small amounts of adaptive data, and 2. Conditionally training the entire model through a speaker embedding of the target speaker. However, both of these methods require adaptive data to appear during training, which makes the training cost to generate new voices quite expensively. In addition, the traditional TTS model uses a simple loss function to reproduce the acoustic features. However, this optimization is based on incorrect distribution assumptions leading to noisy composite audio results. We introduce the Adapt-TTS model that allows high-quality audio synthesis from a small adaptive sample without training to solve these problems. Key recommendations: 1. The Extracting Mel-vector (EMV) architecture allows for a better representation of speaker characteristics and speech style; 2. An improved zero-shot model with a denoising diffusion model (Mel-spectrogram denoiser) component allows for new voice synthesis without training with better quality (less noise). The evaluation results have proven the model's effectiveness when only needing a single utterance (1-3 seconds) of the reference speaker, the synthesis system gave high-quality synthesis results and achieved high similarity.

Vowel classification with combining pitch detection and one‐dimensional convolutional neural network based classifier for gender identification

AbstractHuman speech signals may contain specific information regarding a speaker's characteristics, and these signals can be very useful in applications involving interactive voice response (IVR) and automatic speech recognition (ASR). For IVR and ASR applications, speaker classification into different ages and gender groups can be applied in human–machine interaction or computer‐based interaction systems for customised advertisement, translation (text generation), machine dialog systems, or self‐service applications. Hence, an IVR‐based system dictates that ASR should function through users' voices (specific voice‐frequency bands) to identify customers' age and gender and interact with a host system. In the present study, we intended to combine a pitch detection (PD)‐based extractor and a voice classifier for gender identification. The Yet Another Algorithm for Pitch Tracking (YAAPT)‐based PD method was designed to extract the voice fundamental frequency (F0) from non‐stationary speaker's voice signals, allowing us to achieve gender identification, by distinguishing differences in F0 between adult females and males, and classify voices into adult and children groups. Then, in vowel voice signal classification, a one‐dimensional (1D) convolutional neural network (CNN), consisted of a multi‐round 1D kernel convolutional layer, a 1D pooling process, and a vowel classifier that could preliminary divide feature patterns into three level ranges of F0, including adult and children groups. Consequently, a classifier was used in the classification layer to identify the speakers' gender. The proposed PD‐based extractor and voice classifier could reduce complexity and improve classification efficiency. Acoustic datasets were selected from the Hillenbrand database for experimental tests on 12 vowels classifications, and K‐fold cross‐validations were performed. The experimental results demonstrated that our approach is a very promising method to quantify the proposed classifier's performance in terms of recall (%), precision (%), accuracy (%), and F1 score.

JA PANESE LOANWORDS IN THE CZECH NATIONAL CORPUS

The article discusses some aspects of the functioning of Czech nouns of Japanese origin. Both lexical Japanisms named in the New Encyclopedic Dictionary of the Czech Language (2016) and nouns included in the Neomat electronic database in 2019, 2020 and 2021 are considered. The data of the Czech written and spoken corpora, which are constituents of the Czech National Corpus project, are used. The absolute and relative frequency of use of a particular lexeme, as well as its ARF (= Average Reduced Frequency) was determined with the help of the 4.9 billion-strong corpus of modern written Czech texts SYNv9. The WaG (= Word at a Glance) program developed within the framework of the Czech National Corpus project to determine the regional, gender, age and educational characteristics of Czech speakers using a particular lexeme was used. Russian correspondences to the Czech words of Japanese origin under consideration were given for purely auxiliary purposes, so that the principles of their graphic design were not discussed.

Sinc‐attention feature extraction for trivial‐event based speaker verification

AbstractHuman speech contains trivial events that are non‐subjectively controlled and are limited by physical characteristics. These events are represented by a series of extremely short utterances, such as the ‘hmm’ phrase used for expressing doubt or confirmation. In speaker verification (SV), such events can be a robust choice to verify the real speaker from disguised speech since they are less affected by the randomness of pronunciation. However, trivial events like ‘hmm’ contain little linguistic information and are extremely short, thus the performance of SV systems will decrease drastically. In this letter, a Sinc‐Attention feature extraction method is proposed to extract more discriminative speech features from speech signals to achieve a robust SV system for trivial events. Learnable filters are utilized to obtain low‐level representations of the speech for SV. Moreover, a novel adaptive weighting of features inspired by attention mechanisms is proposed, which effectively improves the representation capability of speaker features. The experiments on different models prove that our method helps SV systems achieve a lower equal error rate (EER) than hand‐crafted feature‐based systems.

Exploring The Speech Patterns of Multilingual Youtubers: an Individual Multilingualism Perspective

Wide exposure to cultural diversity often leads to people learning multiple languages. Accordingly, this factor causes bilingual or multilingual people to experience linguistic phenomena related to changes in the form of spoken and written language they use to communicate with others. Our study sought to find the consequences of being multilingual speakers focused on young YouTubers and identify their speech styles in their YouTube videos. Using a descriptive qualitative method, we collected the data using free-of-conversation listening and note-taking techniques. The result found that multilingual speakers alter their language morphologically and phonologically, affecting their way of speaking. At the morphology level, frequent use of intra-sentential switching and altering grammar and syntax was found. While at the phonology level, the original accents of some YouTubers were often carried away when they spoke in other languages. Also, three factors that affect the language choices of the YouTubers were found, namely participant, function of interaction, and setting. By examining the YouTuber multilingual community, this paper contributes to the development of the scholarship on multilingualism, offering another insight into the linguistic features of multilingual speakers and the way the subjects adjust languages to reach their purposes.
 Keywords: multilingualism, code-switching, language alternation, language choice

Siri, you've changed! Acoustic properties and racialized judgments of voice assistants

As speech technology is increasingly integrated into modern American society, voice assistants are a more significant part of our everyday lives. According to Apple, Siri fulfills 25 billion requests each month. As part of a software update in April 2021, users in the U.S. were presented with a choice of 4 Siris. While in beta testing, users on Twitter began to comment that they felt that some of the voices had racial identities, noting in particular that Voice 2 and Voice 3 “sounded black.” This study tests whether listeners indeed hear the different Siri voices as sounding like speakers from different groups, as well as examines voice quality features that may trigger these judgments. In order to test evaluations of the four voices, 485 American English listeners heard each Siri voice reading the Rainbow Passage, via online survey conducted on Qualtrics. Following each clip, listeners responded to questions about the speaker's demographic characteristics and personal traits. An LMER model of normalized ratings assessed the interaction of voice and race judgment revealed that indeed, Voice 2 and Voice 3 were significantly more likely to be rated as belonging to a Black speaker than Voices 1 and 4 (p < 0.001). Per-trait logistic regression models and chi-square tests examining ratings revealed Voice 3, the male voice rated as Black, was judged less competent (X2 = 108.99, x < 0.001), less professional (X2 = 90.97, p < 0.001), and funniest (X2 = 123.39, x < 0.001). Following analysis of listener judgments of voices, I conducted post-hoc analysis comparing voice quality (VQ) features to examine which may trigger the listener judgments of race. Using PraatSauce, I employed scripts to extract VQ measures previously hypothesized to pattern differently in African American English vs. Mainstream American English. VQ measures that significantly affected listener ratings of the voices are mean F0 and H1–A3c, which correlate with perceptions of pitch and breathiness. These results reveal listeners attribute human-like demographic and personal characteristics to synthesized voices. A more comprehensive understanding of social judgments of digitized voices may help us to understand how listeners evaluate human voices, with implications for speech perception and discrimination as well as recognition and synthesis.

SUNET: Speaker-utterance interaction Graph Neural Network for Emotion Recognition in Conversations

Emotion Recognition in Conversations (ERC) can capture the speakers’ emotional changes in multiple rounds of conversation, so it has a wide range of applications. In recent years, Graph Neural Networks have been naturally used in ERC tasks due to their ability to capture complex non-Euclidian spatial features. However, how to model conversations easily and effectively to improve the effect of ERC in the complex interaction mode still needs to be explored. To this end, we propose a new approach to construct a speaker-utterance interactive heterogeneous network that effectively models context while taking into account the global characteristics of speakers. On this basis, we propose a graph neural network based on the speaker and the corresponding utterances interactions, which dynamically updates the representations of utterances and speakers according to the order in which the speakers talk. We formulate different update methods for utterance and speaker nodes to ensure that the particularity of the heterogeneous network is fully explored. We conduct extensive experiments on four ERC benchmark datasets, and our approach achieves an average 0.7% performance improvement over the most advanced methods, which validates the effectiveness of properly modeling speakers in ERC tasks.

Few-shot short utterance speaker verification using meta-learning.

Short utterance speaker verification (SV) in the actual application is the task of accepting or rejecting the identity claim of a speaker based on a few enrollment utterances. Traditional methods have used deep neural networks to extract speaker representations for verification. Recently, several meta-learning approaches have learned a deep distance metric to distinguish speakers within meta-tasks. Among them, a prototypical network learns a metric space that may be used to compute the distance to the prototype center of speakers, in order to classify speaker identity. We use emphasized channel attention, propagation and aggregation in TDNN (ECAPA-TDNN) to implement the necessary function for the prototypical network, which is a nonlinear mapping from the input space to the metric space for either few-shot SV task. In addition, optimizing only for speakers in given meta-tasks cannot be sufficient to learn distinctive speaker features. Thus, we used an episodic training strategy, in which the classes of the support and query sets correspond to the classes of the entire training set, further improving the model performance. The proposed model outperforms comparison models on the VoxCeleb1 dataset and has a wide range of practical applications.

Sweep-to-Unlock: Fingerprinting Smartphones Based on Loudspeaker Roll-Off Characteristics

Fingerprinting smartphones based on acoustic characteristics of their loudspeaker may have a number of applications in device-to-device authentication as well as in forensic investigations. In this work we propose an efficient fingerprinting methodology by using the roll-off characteristics of the device speaker, i.e., the transition between the low and high stopbands to the passband segment of the speaker. We extract roll-off characteristics from sweep signals, also know as chirps, that are commonly used in practice to test speaker response. This procedure appears to be more stable against variations of the volume level and allows the use of simple linear approximations, which are intuitive and easy to compute, in order to extract the fingerprint. To increase detection accuracy, on the basis of the proven performance of deep learning techniques, a convolutional and a bi-directional long short term memory neural network are further proposed and their performance demonstrated for authentication purposes. While numerous applications may be envisioned, we specifically focus on the use of speaker characteristics in relation to in-vehicle infotainment units, checking if recordings from these units can be used to fingerprint a specific phone.

You're Not You When You're Angry: Robust Emotion Features Emerge by Recognizing Speakers

The robustness of an acoustic emotion recognition system hinges on first having access to features that represent an acoustic input signal. These representations should abstract extraneous low-level variations present in acoustic signals and only capture speaker characteristics relevant for emotion recognition. Previous research has demonstrated that, in other classification tasks, when large labeled datasets are available, neural networks trained on these data learn to extract robust features from the input signal. However, the datasets used for developing emotion recognition systems remain significantly smaller than those used for developing other speech systems. Thus, acoustic emotion recognition systems remain in need of robust feature representations. In this work, we study the utility of speaker embeddings, representations extracted from a trained speaker recognition network, as robust features for detecting emotions. We first study the relationship between emotions and speaker embeddings, and demonstrate how speaker embeddings highlight the differences that exist between neutral speech and emotionally expressive speech. We quantify the modulations that variations in emotional expression incur on speaker embeddings, and show how these modulations are greater than those incurred from lexical variations in an utterance. Finally, we demonstrate how speaker embeddings can be used as a replacement for traditional low-level acoustic features for emotion recognition.

Enhancement in Speaker Identification through Feature Fusion using Advanced Dilated Convolution Neural Network

There are various challenges in identifying the speakers accurately. The Extraction of discriminative features is a vital task for accurate identification in the speaker identification task. Nowadays, speaker identification is widely investigated using deep learning. The complex and noisy speech data affects the performance of Mel Frequency Cepstral Coefficients (MFCC); hence, MFCC fails to represent the speaker characteristics accurately. In this proposed work, a novel text-independent speaker identification system is developed to enhance the performance by fusion of Log-MelSpectrum and excitation features. The excitation information is obtained due to the vibration of vocal folds, and it is represented using Linear Prediction (LP) residual. The various types of features extracted from the excitation are residual phase, sharpness, Energy of Excitation (EoE), and Strength of Excitation (SoE). The extracted features were processed with the dilated convolution neural network (dilated CNN) to fulfill the identification task. The extensive evaluation showed that the fusion of excitation features gives better results than the existing methods. The accuracy reaches 94.12% for 11 complex classes and 91.34% for 80 speakers, and Equal Error Rate (EER) is reduced to 1.16% for the proposed model. The proposed model is tested with the Librispeech corpus using Matlab 2021b tool, outperforming the existing baseline models. The proposed model achieves an accuracy improvement of 1.34% compared to the baseline system.

Ассоциативно-вербальная сеть языка PLAUTDIETSCH немцев-меннонитов Сибири

The paper examines the language of the Mennonite Germans Plautdietsch from a psycholinguistic view point. First, the article describes the history of the Mennonite Germans, discusses the problem of determining their linguistic status in the dialect-Mundart- language paradigm and singles out Plautdietsch as an independent language. Further, a brief description of Mennonite lifestyle is given, in particular, that of the inhabitants of Neudachino village in Novosibirsk region, where the research material was collected. Due to the absence of an established written tradition, it was not possible to conduct the associative experiment in written form, both stimuli and reactions were recorded and then transcribed in Cyrillic, the spelling understandable to the informants. Based on the results of a pilot associative experiment conducted with 13 Plautdietsch speakers in 2021, syntagmatic and paradigmatic relations of stimuli and reactions were studied. By calculating the total brightness index, the core, semi-periphery, and periphery of Mennonite Germans’ linguistic consciousness have been distinguished. According to the semantics of the associations, a verbal-associative network that consists of five thematic groups and expresses the linguistic consciousness of Mennonite Germans (Everyday Life, Possessions, Space, Time and Conceptual Categories) was formed. Already at the stage of the pilot experiment, the conducted research allows us to draw conclusions about the uniqueness and autonomy of Plautdietsch, as well as to highlight the key features of the native speakers’ linguistic worldview.

Phonic Convergence in Estonian Yiddish

Abstract Multilingualism is a characteristic feature of Yiddish speakers in the Baltic region, and contact-induced language change is natural in this case. To date, some Baltic German impact on the local varieties of Yiddish has been discussed in the literature. This article focuses on the phonic impact of Estonian on Estonian Yiddish. The study is based on acoustic analysis of Estonian Yiddish sound recordings. The North Baltic area forms a separate region in the Baltic Sprachbund; Yiddish spoken in Samogitia and Courland as well as Estonian Yiddish exhibit several features typical of this language area. It is demonstrated that there is an overlap of basic prosodic categories with co-territorial languages, such as centralized stress and a quantity opposition. A distinctive feature common to Estonian and Estonian Yiddish is the limited reduction of vowels and a tendency to maintain stable durational ratios of the primary stressed syllable and following unstressed syllables.

Afghan English Major Students’ Attitudes Toward Native-Like Proficiency in the English Language

The term native-like proficiency in English refers to the ability to speak the English language like a native speaker. This study was conducted to investigate Afghan English major undergraduate students’ attitudes toward native-like proficiency in the English language at two public universities in Afghanistan. The study specifically investigated the attitudes of Afghan English major undergraduate students toward the characteristics of a native speaker of the English language, native speakers’ linguistic imperialism, and native-like proficiency in the English language. A mixed research design was used for the study. The quantitative data were descriptively analyzed in terms of frequencies using SPSS and presented in tables. The qualitative data were thematically analyzed and presented. A total of 113 fourth-year (senior students) voluntarily participated in the study. The findings revealed that perfect pronunciation, fluency, and grammatical accuracy were the defining attributes of native speakers. Similarly, students believed that only native speakers have the right to decide the rules or to standardize the English language and that only a native speaker has better English language skills. Moreover, the results revealed that Afghan English major students’ attitudes toward native-like proficiency in the English language were highly positive, and they love to speak with a native-like accent. It is hoped that the current study’s findings will help educators and will be a valuable addition to the current mass of data on the subject.