Vowel Recognition Research Articles

Enhancing Mandarin Tone and Vowel Recognition in Cochlear Implant Simulation through Combining Task Practice

Purpose: Mandarin tone recognition with cochlear implant simulation can be enhanced via targeted auditory training. However, such training generally does not transfer to different tasks, even when they share the same stimuli. As such, this study investigated whether combined Mandarin tone and vowel recognition training leads to greater improvements in Mandarin tone, vowel, and sentence recognition compared to tone- or vowel- only training.Methods: Twenty-nine native Mandarin-speaking young adults with normal hearing were randomly assigned to one of three training regimens: Tone-Only, Vowel-Only, Tone-Vowel. Mandarin tone, vowel, and sentence recognition were tested before and after training.Results: Tone-Only training improved Mandarin tone recognition but not vowel recognition, while Vowel-Only training improved Mandarin vowel recognition but not tone recognition. The combined Tone-Vowel training improved both tone and vowel recognition. ToneVowel training also led to improvements in Mandarin sentence recognition, although these gains were not significantly greater than those observed in Tone-Only or Vowel-Only groups.Conclusion: These findings suggest that within an equivalent amount of training, combined tone and vowel recognition training enhances both tone and vowel recognition, while tone- or vowel-only training improves the performance in the targeted task alone. Combined tone and vowel recognition may be necessary to achieve improvements in Mandarin tone, vowel, and sentence recognition.

Vowel Recognition for Rehabilitation Assessment of Speech Disorder Patients via Multi-source Frequency Spectrum Images

هناك مجموعة واسعة من الأسباب الطبية لضعف الاتصال ، مثل اضطرابات الكلام ، وفقدان السمع ، وإصابات الدماغ ، والسكتة الدماغية ، والإعاقات الجسدية. نتيجة لذلك ، يمكن أن يؤثر اضطراب التواصل مدى الحياة على التنمية الاجتماعية والعلاقة الشخصية. يمكن أن تستفيد اضطرابات النطق من علاجات النطق المبكرة ؛ ومع ذلك ، لا تزال غالبية مرافق إعادة التأهيل في جميع أنحاء العالم تنفذ هذه العملية يدويًا. من وجهة نظر عالمية ، تم إجراء مجموعة واسعة من الدراسات حول معالجة الكلام لمختلف اللغات البشرية. نظرًا لأن رؤية الكمبيوتر قد أثرت على هذا المجال ، فقد تم تطبيق التعلم الآلي والتعلم العميق في الصناعة الطبية والرعاية الصحية لتعزيز إعادة التأهيل من خلال استخدام التكنولوجيا الجديدة. حللت هذه الدراسة دقة تصنيف الشبكة المصممة والنماذج الأخرى المدربة مسبقًا (VGG-Net و AlexNet و Inception) وأجرت تحليلًا مقارنًا كاملًا لتقييم دقة التصنيف للعديد من النماذج المدربة مسبقًا. في هذا العمل المقترح ، لإنجاز مهمة التصنيف هذه ، يتم تحويل الصوت لاحقًا إلى الصورة كطريقة جديدة لرؤيتها في الشبكة العصبية عبر مفهوم مقترح حديثًا يسمى بيانات ملف تعريف الصورة. أنتجت مجموعات البيانات التي تم تصنيفها عن طريق الصور والتي استخدمت مخططًا طيفيًا ومعامل تردد ميل التردد (MFCC) أفضل نتائج هذه الدراسة ودقتها. يهدف هذا المشروع إلى تطوير شبكة عصبية جديدة يمكنها التمييز بنجاح بين أحرف العلة من أصوات الأشخاص العاديين والمرضى الذين يعانون من اضطرابات الكلام والمزيج من المجموعتين السابقتين باستخدام الفئتين الستة والثاني عشر من حروف العلة الملايو. وفقًا للبيانات التجريبية التي تم إجراؤها ، ونموذج الشبكة المصمم ، والذي استخدم 6 أحجام دفعات ، و 20 حقبة ، و ADAM كمحسِّن ، قدم هذا المشروع وحقق قيم الدقة القصوى لكلا الفئتين لبيانات الصوت الخاصة بالصور في جميع التحليلات التي تم إجراؤها.

Enhancing Speech Impairment Support: Designing an EEG-Based BCI System for Turkish Vowel Recognition

Enhancing Speech Impairment Support: Designing an EEG-Based BCI System for Turkish Vowel Recognition

The impact of spectral and temporal processing on speech recognition in children with cochlear implants

While the relationships between spectral resolution, temporal resolution, and speech recognition are well defined in adults with cochlear implants (CIs), they are not well defined for prelingually deafened children with CIs, for whom language development is ongoing. This cross-sectional study aimed to better characterize these relationships in a large cohort of prelingually deafened children with CIs (N = 47; mean age = 8.33 years) by comprehensively measuring spectral resolution thresholds (measured via spectral modulation detection), temporal resolution thresholds (measured via sinusoidal amplitude modulation detection), and speech recognition (measured via monosyllabic word recognition, vowel recognition, and sentence recognition in noise via both fixed signal-to-noise ratio (SNR) and adaptively varied SNR). Results indicated that neither spectral or temporal resolution were significantly correlated with speech recognition in quiet or noise for children with CIs. Both age and CI experience had a moderate effect on spectral resolution, with significant effects for spectral modulation detection at a modulation rate of 0.5 cyc/oct, suggesting spectral resolution may improve with maturation. Thus, it is possible we may see an emerging relationship between spectral resolution and speech perception over time for children with CIs. While further investigation into this relationship is warranted, these findings demonstrate the need for new investigations to uncover ways of improving spectral resolution for children with CIs.

Vowel Recognition Revolutionizes Early Childhood Education Globally

This study addresses the challenge of vowel recognition among 3-4-year-old children, focusing on vowels like “e,” “u,” and “o.” Using a Classroom Action Research (CAR) approach, twelve children at KB Aisyiyah 17 Jasem participated in pre-cycle, cycle I, and cycle II interventions. The results showed a significant improvement from a 33.3% recognition rate in the pre-cycle to 95.3% in cycle II, indicating the effectiveness of using ball letter media. This research highlights the importance of tailored teaching methods in enhancing early childhood learning outcomes, particularly in improving vowel recognition skills. Highlight: Tailored CAR approach boosts vowel recognition in early education. Effective ball letter media enhances learning outcomes in young children. Notable 62% increase shows success in improving vowel recognition skills. Keyword: Vowel recognition, Early childhood education, Classroom Action Research, Ball letter media, Learning outcomes

English vowel recognition in multi-talker babbles mixed with different numbers of talkersa).

The present study examined English vowel recognition in multi-talker babbles (MTBs) in 20 normal-hearing, native-English-speaking adult listeners. Twelve vowels, embedded in the h-V-d structure, were presented in MTBs consisting of 1, 2, 4, 6, 8, 10, and 12 talkers (numbers of talkers [N]) and a speech-shaped noise at signal-to-noise ratios of -12, -6, and 0 dB. Results showed that vowel recognition performance was a non-monotonic function of N when signal-to-noise ratios were less favorable. The masking effects of MTBs on vowel recognition were most similar to consonant recognition but less so to word and sentence recognition reported in previous studies.

The influence of the stimulus level used to prescribe nonlinear frequency compression on speech perception.

Nonlinear frequency compression (NFC) is a signal processing technique designed to lower high frequency inaudible sounds for a listener to a lower frequency that is audible. Because the maximum frequency that is audible to a listener with hearing loss will vary with the input speech level, the input level used to set nonlinear frequency compression could impact speech recognition. The purpose of this study was to determine the influence of the input level used to set nonlinear frequency compression on nonsense syllable recognition. Nonsense syllable recognition was measured for three NFC fitting condition (i.e., with nonlinear frequency compression set based on speech input levels of 50-, 60-, and 70-dB SPL, respectively), as well as without nonlinear frequency compression (restricted bandwidth condition). Twenty-three adults (ages 42-80 years old) with hearing loss. Data were collected, monaurally, using a hearing aid simulator. The start frequency and frequency compression ratios were set based on the SoundRecover Fitting Assistant. Speech stimuli were 657 consonant-vowel-consonant nonwords presented at 50, 60, and 70 dB SPL and mixed with steady noise (6 dB SNR) and scored based on entire word, initial consonant, vowel, and final consonant. Linear mixed effects examined the effects of NFC fitting condition , presentation level, and scoring method on percent correct recognition. Additional predictor variables of start frequency and frequency-compression ratio were examined. Nonsense syllable recognition increased as presentation level increased. Nonsense syllable recognition for all presentation levels was highest when nonlinear frequency compression was set based on the 70 dB SPL input level and decreased significantly when set based on the 60- and 50-dB SPL inputs. Relative to consonant recognition, there was a greater reduction in vowel recognition. Nonsense syllable recognition between NFC fitting conditions improved with increases in the start frequency, where higher start frequencies led to better nonsense word recognition. Nonsense syllable recognition was highest when setting nonlinear frequency compression based on a 70 dB SPL presentation level and suggest that a high presentation level should be used to determine nonlinear frequency compression parameters for an individual patient.

Percepcija samoglasniških opozicij pri dijakih pomurskih srednjih šol

The article presents the reception and recognition of vowels. A survey in form of a perception test was conducted among the pupils of four high schools in the Pomurje region, i.e. Murska Sobota, Ljutomer and Lendava. The obtained results were analysed and compared with the research carried out among the students of Slavonic and Slovene studies (Tivadar 2004a). The study shows that pupils are not doing well at distinguishing vowel quality, while the length of individual vowels hardly ever influences the distinction between the semantic properties of words. Of all subjects involved in the study, the pupils from the Lendava grammar school had most difficulties with vowel recognition, while it is also interesting to note that vowels were better recognised by female pupils (by as much as 17 %).

Vowel’s classification for stroke patients through rehabilitation performance via image-profiled sound data

In terms of medicine, a disorder is a disturbance of the mind or body's normal functioning. Communication issues may result from stroke because it damages the parts of the brain that control language. The capacity to talk, read, write, and comprehend speeches can all be affected after a stroke. The rehabilitation and treatment of patients generally take a long time and include constant medication, exercise, and rehabilitation training. However, most rehab facilities throughout the world still manually carry out this rehabilitation process. Machine learning and deep learning have been introduced to this medical field to aid rehabilitation using the new technology due to computer vision's impact on this field. A reliable Convolution Neural Network with a graphical user interface is introduced in this study to support and enhance rehabilitation efforts. The spectrogram in the image-profiled sound is used to provide the optimum outcome and accuracy. This project aims to develop a neural network that can distinguish vowels between a normal person's and stroke patients' voices. In this proposed paper's result, an intelligent Convolution Neural Network system for Malay language vowel detection with high-performance accuracy is demonstrated with maximum accuracy was 92.96% with 20 epoch numbers and 6 batch size. This outcome showed that the proposed method, even using a simple network design, is still competitive compared with other methods. The proposed method is ideal for training and validating vowel recognition accuracy especially for stroke patients.

Comparison of objective and subjective methods for evaluating speech quality and intelligibility recorded through bone conduction and in-ear microphones

Bone conduction and in-ear microphones pick up the Bone-conducted (BC) speech, which has traveled through bones and soft tissues. This BC speech is less sensitive to surrounding noise than Air-conducted (AC) speech. Therefore, there is an interest in recording this signal. However, the intelligibility and quality of these microphones are known to be a limit to their use. Previous work has noted confusion between vowel sounds in terms of intelligibility. To evaluate the quality and intelligibility, studies rely on subjective and objective tests. This paper aims to determine if the standard objective methods for rating speech quality and intelligibility can be applied to BC speech recorded through these microphones. In order to estimate intelligibility, a subjective test based on vowel recognition was compared to STOI and to a new criterion based on second formant frequencies of oral vowels. For speech quality estimation, MUSHRA and PESQ tests were compared. Results show difficulties in using objective methods instead of subjective ones. Therefore, for ongoing studies, it is suggested to use subjective methods to evaluate speech quality and intelligibility.

A Study on the Impact of Lombard Effect on Recognition of Hindi Syllabic Units Using CNN Based Multimodal ASR Systems

Research work on the design of robust multimodal speech recognition systems making use of acoustic and visual cues, extracted using the relatively noise robust alternate speech sensors is gaining interest in recent times among the speech processing research fraternity. The primary objective of this work is to study the exclusive influence of Lombard effect on the automatic recognition of the confusable syllabic consonant-vowel units of Hindi language, as a step towards building robust multimodal ASR systems in adverse environments in the context of Indian languages which are syllabic in nature. The dataset for this work comprises the confusable 145 consonant-vowel (CV) syllabic units of Hindi language recorded simultaneously using three modalities that capture the acoustic and visual speech cues, namely normal acoustic microphone (NM), throat microphone (TM) and a camera that captures the associated lip movements. The Lombard effect is induced by feeding crowd noise into the speaker’s headphone while recording. Convolutional Neural Network (CNN) models are built to categorise the CV units based on their place of articulation (POA), manner of articulation (MOA), and vowels (under clean and Lombard conditions). For validation purpose, corresponding Hidden Markov Models (HMM) are also built and tested. Unimodal Automatic Speech Recognition (ASR) systems built using each of the three speech cues from Lombard speech show a loss in recognition of MOA and vowels while POA gets a boost in all the systems due to Lombard effect. Combining the three complimentary speech cues to build bimodal and trimodal ASR systems shows that the recognition loss due to Lombard effect for MOA and vowels reduces compared to the unimodal systems, while the POA recognition is still better due to Lombard effect. A bimodal system is proposed using only alternate acoustic and visual cues which gives a better discrimination of the place and manner of articulation than even standard ASR system. Among the multimodal ASR systems studied, the proposed trimodal system based on Lombard speech gives the best recognition accuracy of 98%, 95%, and 76% for the vowels, MOA and POA, respectively, with an average improvement of 36% over the unimodal ASR systems and 9% improvement over the bimodal ASR systems.

A novel negative selection algorithm with hyper-rectangle detectors based on full coverage of state space for anomaly detection

The traditional negative selection algorithm with a randomly generated hypersphere detector is unable to satisfy the development needs of continuous learning due to the inherent defects of the detector. This paper proposes a novel negative selection algorithm for hyper-rectangle detectors that overcomes the shortcomings of randomly generated hyper-sphere detectors and lays the foundation for a negative selection algorithm with continuous learning capability. It uses self-sample clusters of equal-sized hypercubes instead of self-samples for training. The hyper-rectangle detectors are generated by cutting the nonself-space along the boundary of the self-sample clusters. The state space is covered without overlapping each other by self-sample clusters and detectors. The anomaly detection performance of the proposed method was demonstrated using Iris data, vowel recognition data (Vowel), and Wisconsin Breast Cancer (BCW) data. The experimental results show that the proposed method outperforms other artificial immune algorithms and clustering algorithms under the same parameter conditions.

Influence of the Frequency-to-Place Function on Recognition with Place-Based Cochlear Implant Maps.

Comparison of acute speech recognition for cochlear implant (CI) alone and electric-acoustic stimulation (EAS) users listening with default maps or place-based maps using either a spiral ganglion (SG) or a new Synchrotron Radiation-Artificial Intelligence (SR-AI) frequency-to-place function. Thirteen adult CI-alone or EAS users completed a task of speech recognition at initial device activation with maps that differed in the electric filter frequency assignments. The three map conditions were: (1) maps with the default filter settings (default map), (2) place-based maps with filters aligned to cochlear SG tonotopicity using the SG function (SG place-based map), and (3) place-based maps with filters aligned to cochlear Organ of Corti (OC) tonotopicity using the SR-AI function (SR-AI place-based map). Speech recognition was evaluated using a vowel recognition task. Performance was scored as the percent correct for formant 1 recognition due to the rationale that the maps would deviate the most in the estimated cochlear place frequency for low frequencies. On average, participants had better performance with the OC SR-AI place-based map as compared to the SG place-based map and the default map. A larger performance benefit was observed for EAS users than for CI-alone users. These pilot data suggest that EAS and CI-alone users may experience better performance with a patient-centered mapping approach that accounts for the variability in cochlear morphology (OC SR-AI frequency-to-place function) in the individualization of the electric filter frequencies (place-based mapping procedure). 3 Laryngoscope, 133:3540-3547, 2023.

Dopant network processing units as tuneable extreme learning machines

Inspired by the highly efficient information processing of the brain, which is based on the chemistry and physics of biological tissue, any material system and its physical properties could in principle be exploited for computation. However, it is not always obvious how to use a material system’s computational potential to the fullest. Here, we operate a dopant network processing unit (DNPU) as a tuneable extreme learning machine (ELM) and combine the principles of artificial evolution and ELM to optimise its computational performance on a non-linear classification benchmark task. We find that, for this task, there is an optimal, hybrid operation mode (“tuneable ELM mode”) in between the traditional ELM computing regime with a fixed DNPU and linearly weighted outputs (“fixed-ELM mode”) and the regime where the outputs of the non-linear system are directly tuned to generate the desired output (“direct-output mode”). We show that the tuneable ELM mode reduces the number of parameters needed to perform a formant-based vowel recognition benchmark task. Our results emphasise the power of analog in-matter computing and underline the importance of designing specialised material systems to optimally utilise their physical properties for computation.

Vowel focused training improves listener transcription of dysarthric speech

Typical listeners can adjust to the speech of individuals with dysarthria through the process of perceptual learning. Past research has demonstrated that listeners improve in their recognition of both segments and connected speech produced by people with dysarthria. The mechanisms underlying learning of dysarthric speech are still uncertain, though it has been suggested that exposure allows listeners to tune into the segmental characteristics of speech. In the current study, we test this hypothesis by training listeners to identify vowels spoken by an individual with mild dysarthria. We employed a pre-test/post-test design where we first tested listeners on (1) vowel recognition and (2) transcription accuracy of connected speech, trained them, and then tested them again. We found that listeners who were trained on dysarthric vowels demonstrated greater improvements in vowel identification than those in the control condition. Likewise, the listeners who underwent training showed greater improvements in transcription accuracy than those in the control condition. However, the training advantage did not generalize to unfamiliar phrases. Overall, it appears that listeners are able to tune into the segmental characteristics of dysarthric speech after a short training session, which improves their recognition of longer connected speech.

Comparable Encoding, Comparable Perceptual Pattern: Acoustic and Electric Hearing.

Perception with electric neuroprostheses is sometimes expected to be simulated using properly designed physical stimuli. Here, we examined a new acoustic vocoder model for electric hearing with cochlear implants (CIs) and hypothesized that comparable speech encoding can lead to comparable perceptual patterns for CI and normal hearing (NH) listeners. Speech signals were encoded using FFT-based signal processing stages including band-pass filtering, temporal envelope extraction, maxima selection, and amplitude compression and quantization. These stages were specifically implemented in the same manner by an Advanced Combination Encoder (ACE) strategy in CI processors and Gaussian-enveloped Tones (GET) or Noise (GEN) vocoders for NH. Adaptive speech reception thresholds (SRTs) in noise were measured using four Mandarin sentence corpora. Initial consonant (11 monosyllables) and final vowel (20 monosyllables) recognition were also measured. Naïve NH listeners were tested using vocoded speech with the proposed GET/GEN vocoders as well as conventional vocoders (controls). Experienced CI listeners were tested using their daily-used processors. Results showed that: 1) there was a significant training effect on GET vocoded speech perception; 2) the GEN vocoded scores (SRTs with four corpora and consonant and vowel recognition scores) as well as the phoneme-level confusion pattern matched with the CI scores better than controls. The findings suggest that the same signal encoding implementations may lead to similar perceptual patterns simultaneously in multiple perception tasks. This study highlights the importance of faithfully replicating all signal processing stages in the modeling of perceptual patterns in sensory neuroprostheses. This approach has the potential to enhance our understanding of CI perception and accelerate the engineering of prosthetic interventions. The GET/GEN MATLAB program is freely available at https://github.com/BetterCI/GETVocoder.

IR-UWB Radar-Based Contactless Silent Speech Recognition of Vowels, Consonants, Words, and Phrases

IR-UWB Radar-Based Contactless Silent Speech Recognition of Vowels, Consonants, Words, and Phrases

Memory-card vowel training for child and adult second-language learners: A first report.

Japanese adults and Spanish-Catalan children received auditory phonetic training for English vowels using a novel paradigm, a version of the common children's card game Concentration. Individuals played a computer-based game in which they turned over pairs of cards to match spoken words, drawn from sets of vowel minimal pairs. The training was effective for adults, improving vowel recognition in a game that did not explicitly require identification. Children likewise improved over time on the memory card game, but not on the present generalisation task. This gamified training method can serve as a platform for examining development and perceptual learning.

تماثلية الحركات في العربية، وفي ما يناظرها في مربع جونز على ضوء تماثلية مادسون: دراسة تحليلية

Acoustic phonetic studies have become fashionable in the past two decades. Many theories have been advanced to analyse sounds experimentally. Vowel analysis gained traction after Jones' Cardinal vowels recognition. Vowel spacing and dispersion in Arabic in the light of what Jones has carried out and the relatively more recent theory of dispersion expounded by Lan Maddieson, has been thoroughly and meticulously examined. Lan Maddieson views that vowels are evenly spaced and dispersed within the designated oral space. The triangular shape and dispersion of the three Arabic vowels was compared to that of Jones' Cardinal vowels numbered 1، 5 and 8. The two vocalic systems seem to share striking similarities with regard to distribution and dispersion. 85% compatibility and correspondence was observed، which lends support to Maddieson's theory.

Silent Speech Recognition with Strain Sensors and Deep Learning Analysis of Directional Facial Muscle Movement.

Silent communication based on biosignals from facial muscle requires accurate detection of its directional movement and thus optimally positioning minimum numbers of sensors for higher accuracy of speech recognition with a minimal person-to-person variation. So far, previous approaches based on electromyogram or pressure sensors are ineffective in detecting the directional movement of facial muscles. Therefore, in this study, high-performance strain sensors are used for separately detecting x- and y-axis strain. Directional strain distribution data of facial muscle is obtained by applying three-dimensional digital image correlation. Deep learning analysis is utilized for identifying optimal positions of directional strain sensors. The recognition system with four directional strain sensors conformably attached to the face shows silent vowel recognition with 85.24% accuracy and even 76.95% for completely nonobserved subjects. These results show that detection of the directional strain distribution at the optimal facial points will be the key enabling technology for highly accurate silent speech recognition.