Vowel Recognition Research Articles

Matching Acoustical Properties and Native Perceptual Assessments of L2 Speech

AbstractThis article analyses the acoustical properties of Dutch vowels produced by adult Spanish learners and investigates how these vowels are perceived by non-expert native Dutch listeners. Statistical vowel classifications obtained from the acoustical properties of the learner vowel realizations were compared to vowel classifications provided by native Dutch listeners. Both types of classifications were affected by the specific set of vowels included as stimuli, an effect caused by the large variability in Spanish learners’ vowel realizations. While there were matches between the two types of classifications, shifts were noted within and between production and perception, depending on the vowel and vowel features. We considered the variability between Spanish learners further by investigating individual patterns in the production and perception data, and linking these to the learners’ proficiency level and multilingual background. We conclude that integrating production and perception data provides valuable insights into the role of different features in adult L2 learning, and how their properties actively interact in the way L2 speech is perceived. A second conclusion is that adaptive mechanisms, signalled by boundary shifts and useful in coping with variability of non-native vowel stimuli, play a role in both statistical vowel classifications (production) and human vowel recognition (perception).

Vowel Instruction for Spelling Enhancement of Arabic-Speaking Learners of English

Word spellings are often a common linguistic issue among Arabic-speaking learners of English, and English vowels are one of the most difficult to resolve. Therefore, the study presented vowel instruction as an intervention to address this issue by determining its effect on the spelling performance on monosyllables and bi-syllables. Vowel instruction sessions were then offered to 15 university freshman students for two months. Instructional strategies consisted of teaching vowel pronunciation, vowel letter-to-sound relationships, vowel spelling patterns, word-breaking, and providing various spelling opportunities. To measure overall spelling performance, pre-test and post-test on vowel spellings were conducted which revealed a significant increase in the mean scores. This suggests that vowel instruction was facilitative in the improved spelling performance on monosyllabic and bi-syllabic words, with more misspellings in the latter than in the former. The strategies used were generally perceived as beneficial, and it is recommended that including them in language lessons could improve vowel spelling skill and minimize English vowel confusions. Moreover, constant exposure of learners to many English words and more spelling activities are still necessary for better English vowel recognition and whole word spelling.

Recognition of indigenous and foreign speech in a noisy classroom: A case of primary school pupils

This study examines the level of speech recognition of English and Igbo utterances by 70 grade four children. The children, whose mother tongue is Igbo and aged between 8 and 10 years, had English monosyllabic words as well as Igbo monosyllabic and disyllabic words dictated to them in noisy and quiet classrooms. The results show that in noise, their level of recognition was significantly higher for English utterances (p < 0.05) while in quiet, it was significantly higher for Igbo (p < 0.05). Thus, the findings confirm those of previous studies that the recognition of Igbo is more affected by noise. In terms of the recognition of phonemes in quiet and noise, for both languages, the ‘t’ test analyses showed significant difference (p < 0.05) in the pupils’ recognition of vowels and consonants in both languages, with vowels being more identified than consonants in both noise and quiet. Hence, though the recognition of both languages is hindered in noise, the study of Igbo requires an acoustically serene environment for maximum results.

Neural network and GMM based feature mappings for consonant–vowel recognition in emotional environment

In this work, we propose a mapping function based feature transformation framework for developing consonant–vowel (CV) recognition system in the emotional environment. An effective way of conveying messages is by expressing emotions during human conversations. The characteristics of CV units differ from one emotion to other emotions. The performance of existing CV recognition systems is degraded in emotional environments. Therefore, we have proposed mapping functions based on artificial neural network and GMM models for increasing the accuracy of CV recognition in the emotional environment. The CV recognition system has been explored to transform emotional features to neutral features using proposed mapping functions at CV and phone levels to minimize mismatch between training and testing environments. Vowel onset and offset points have been used to identify vowel, consonant and transition segments. Transition segments are identified by considering initial 15% speech samples between vowel onset and offset points. The average performance of CV recognition system is increased significantly using feature mapping technique at phone level in three emotional environments (anger, happiness, and sadness).

Handwritten recognition of Tamil vowels using deep learning

We come across a large volume of handwritten texts in our daily lives and handwritten character recognition has long been an important area of research in pattern recognition. The complexity of the task varies among different languages and it so happens largely due to the similarity between characters, distinct shapes and number of characters which are all language-specific properties. There have been numerous works on character recognition of English alphabets and with laudable success, but regional languages have not been dealt with very frequently and with similar accuracies. In this paper, we explored the performance of Deep Belief Networks in the classification of Handwritten Tamil vowels, and conclusively compared the results obtained. The proposed method has shown satisfactory recognition accuracy in light of difficulties faced with regional languages such as similarity between characters and minute nuances that differentiate them. We can further extend this to all the Tamil characters.

An analysis of the effect of combining standard and alternate sensor signals on recognition of syllabic units for multimodal speech recognition

This paper studies the effect of combining evidences from multiple modes of speech on the recognition of different categories of sounds. Multimodal speech recognition systems are built by combining the acoustic and visual cues from the (lip radiated) normal microphone speech, throat microphone speech and lip reading for the recognition of the highly confusable 145 consonant-vowel units of the Hindi language. The performance of the multimodal systems are compared with that of the unimodal systems for the recognition of sounds based on their place (POA) and manner of articulation (MOA) as well as their associated vowels. This comparison shows that though the multimodal ASR systems rely on the presence of complimentary speech-related acoustic and visual cues present in the different modes, not all evidences are complimentary. Bimodal systems that combines visual cues from lip reading are shown to improve the recognition of sounds based on POA and MOA, but decrease the recognition of vowels. This study shows that, compared to the standard Automatic Speech Recognition(ASR) system, the best multimodal system that combines the two acoustic cues as well as visual cue improves the recognition of POA category by 11%, MOA category by 3% and vowels by 2%. However, the study shows the need for exploring better fusion techniques to overcome absence of complementary evidences in certain categories of sounds, especially in bimodal systems.

Integration of acoustic and electric hearing is better in the same ear than across ears

Advances in cochlear implant (CI) technology allow for acoustic and electric hearing to be combined within the same ear (electric-acoustic stimulation, or EAS) and/or across ears (bimodal listening). Integration efficiency (IE; the ratio between observed and predicted performance for acoustic-electric hearing) can be used to estimate how well acoustic and electric hearing are combined. The goal of this study was to evaluate factors that affect IE in EAS and bimodal listening. Vowel recognition was measured in normal-hearing subjects listening to simulations of unimodal, EAS, and bimodal listening. The input/output frequency range for acoustic hearing was 0.1–0.6 kHz. For CI simulations, the output frequency range was 1.2–8.0 kHz to simulate a shallow insertion depth and the input frequency range was varied to provide increasing amounts of speech information and tonotopic mismatch. Performance was best when acoustic and electric hearing was combined in the same ear. IE was significantly better for EAS than for bimodal listening; IE was sensitive to tonotopic mismatch for EAS, but not for bimodal listening. These simulation results suggest acoustic and electric hearing may be more effectively and efficiently combined within rather than across ears, and that tonotopic mismatch should be minimized to maximize the benefit of acoustic-electric hearing, especially for EAS.

Effects of nonlinear frequency compression on Mandarin speech recognition and sound quality perception in hearing-aid users

The aim of the present study is to evaluate the effects of nonlinear frequency compression (NLFC) on Mandarin speech recognition and sound quality perception. Thirty Chinese-speaking, hearing-impaired adults without hearing-aid experiences participated in the study. They were fitted bilaterally with the Phonak behind-the-ear hearing aids. Each participant was evaluated with the Phoneme Perception Test, Mandarin speech recognition, and sound quality perception at post-fitting 0, 4, and 12 weeks with NLFC-on vs. NLFC-off. The NLFC were active at home. Results show significant differences in detection of phonemes of high-frequency components (such as /sa/) between NLFC on and off. The difference between NLFC on and off was significant for consonant and sentence recognition. Duration of NLFC use had significant effects on the recognition of consonants, vowels, and sentences. Regarding the sound quality perception test, there were significant effects of NLFC conditions (on vs. off), duration of NLFC use, and sound types on the four categories of percepts (i.e., loudness, clarity, naturalness, and overall preference). Therefore, NLFC improves the audibility of high-frequency speech signals and provides better recognition of consonants and sentences and perception of sound quality. The benefit of NLFC on speech recognition and sound quality perception requires certain amount of acclimatization.

HANDWRITTEN DEVANAGARI VOWEL RECOGNITION USING ARTIFICIAL NEURAL NETWORK

Handwritten character recognition is active and open field of research in the area of pattern recognition. As, there is continuous developments in the hardware i.e. Machine. At the same time, many researchers are working in the field of character recognition from more than last four decays. Handwritten character recognition involves reading of handwritten character and comparing the required one. Human being is doing this task while learning characters in the childhood. But the same task for machine is much complex. This complexity depends on the environment in which the character was written. For the machine the process of reading, understanding and interpretation of handwritten character is difficult task. This research work proposes new approaches for extracting features in context of Handwritten Devanagari Vowels recognition. For classification technique Artificial Network is used. The overall accuracy of recognition of handwritten Devanagari Vowels is % with SVM classifier, % with MLP and it is %with GFF.

Vowel recognition at fundamental frequencies up to 1 kHz reveals point vowels as acoustic landmarks.

The phonological function of vowels can be maintained at fundamental frequencies (fo) up to 880 Hz [Friedrichs, Maurer, and Dellwo (2015). J. Acoust. Soc. Am. 138, EL36-EL42]. Here, the influence of talker variability and multiple response options on vowel recognition at high fos is assessed. The stimuli (n = 264) consisted of eight isolated vowels (/i y e ø ε a o u/) produced by three female native German talkers at 11 fos within a range of 220-1046 Hz. In a closed-set identification task, 21 listeners were presented excised 700-ms vowel nuclei with quasi-flat fo contours and resonance trajectories. The results show that listeners can identify the point vowels /i a u/ at fos up to almost 1 kHz, with a significant decrease for the vowels /y ε/ and a drop to chance level for the vowels /e ø o/ toward the upper fos. Auditory excitation patterns reveal highly differentiable representations for /i a u/ that can be used as landmarks for vowel category perception at high fos. These results suggest that theories of vowel perception based on overall spectral shape will provide a fuller account of vowel perception than those based solely on formant frequency patterns.

Neuromagnetic correlates of voice pitch, vowel type, and speaker size in auditory cortex

Vowel recognition is largely immune to differences in speaker size despite the waveform differences associated with variation in speaker size. This has led to the suggestion that voice pitch and mean formant frequency (MFF) are extracted early in the hierarchy of hearing/speech processing and used to normalize the internal representation of vowel sounds. This paper presents a magnetoencephalographic (MEG) experiment designed to locate and compare neuromagnetic activity associated with voice pitch, MFF and vowel type in human auditory cortex. Sequences of six sustained vowels were used to contrast changes in the three components of vowel perception, and MEG responses to the changes were recorded from 25 participants. A staged procedure was employed to fit the MEG data with a source model having one bilateral pair of dipoles for each component of vowel perception. This dipole model showed that the activity associated with the three perceptual changes was functionally separable; the pitch source was located in Heschl's gyrus (bilaterally), while the vowel-type and formant-frequency sources were located (bilaterally) just behind Heschl's gyrus in planum temporale. The results confirm that vowel normalization begins in auditory cortex at an early point in the hierarchy of speech processing.

Odia Handwritten Vowel Recognition System using Slantlet Transform and Differential Evolution based Functional Link Artificial Neural Network Classifier

One of the challenging problems in optical character recognition (OCR) is the identification of handwritten characters. It has several complexities like missing part, size variation, added noise, media devices used, slant variation etc. The goal of this paper is to develop an effective system for recognition of Odia handwritten vowels using functional link artificial neural network (FLANN) classifier and differential evolution (DE) optimization technique. In this paper the emphasis is on preprocessing, feature extraction, feature reduction and classification with optimization technique. Various preprocessing operations like size normalization mean filtering and canny edge detection methods are carried out on the vowel images to enhance the quality. Slantlet transform (SLT) is applied for feature extraction and the extracted features are further reduced by principal component analysis (PCA) method. FLANN classifier is applied on the reduced feature vector and the weights of the FLANN are optimized with DE evolutionary technique. The system has achieved a good classification accuracy of 91.32% on the dataset.

Bimodal benefits in Mandarin-speaking cochlear implant users with contralateral residual acoustic hearing

Objective: The present study aimed to measure bimodal benefits and probe their underlying mechanisms in Mandarin-speaking cochlear implant (CI) subjects who had contralateral residual acoustic hearing. Design: The subjects recognised words or phonemes from the Mandarin Lexical Neighborhood Test in noise at a 10-dB signal-to-noise ratio (SNR) with acoustic stimulation, electric stimulation or the combined bimodal stimulation. Study sample: Thirteen Mandarin-speaking subjects wore a CI in one ear and had residual acoustic hearing in the contralateral ear. Six of the subjects (5.2–13.0 years) had pre-lingual onset of severe hearing loss, and seven of them (8.6–45.8 years) had post-lingual onset of severe hearing loss. Results: Both groups of subjects produced a significant bimodal benefit in word recognition in noise. Consonants and tones accounted for the bimodal benefit. The bimodal integration efficiency was negatively correlated with the duration of deafness in the implanted ear for vowel recognition but positively correlated with CI or bimodal experience for consonant recognition. Conclusions: The present results support preservation of residual acoustic hearing, early cochlear implantation and continuous use of bimodal hearing for subjects who have significant residual hearing in the non-implanted ear.

“Flat” vowel spectra revisited in vowel synthesis

Some studies of natural and of synthesized vowel sounds indicate “flat” vowel-related spectral envelopes or envelope parts in terms of vowel-related frequency ranges with harmonics equal in amplitude. The present investigation addresses this question in a vowel synthesis experiment in which sounds related to series of harmonics, multiples of 200 Hz in frequency and equal in amplitude, were created. Thereby, for various frequency ranges, the number of harmonics was increased stepwise from a single lower harmonic to an increasingly broader harmonic series, and, inversely, it was also decreased from a broad series of harmonics to a single higher harmonic. The entire frequency range of investigation was 0.2-4 kHz. Vowel recognition was investigated by means of a listening test in which five phonetic expert listeners were asked to assign the synthesized sounds to Standard German vowel qualities. The results of the experiment reveal that synthesized sounds with frequency bands of series of two or more equal-amplitude harmonics allow for a perceptual differentiation of the Standard German vowels /i-y-e-ɛ-a-ɔ-o/. Methodological issues concerning future investigations as well as implications for the acoustics and perception of vowels are discussed.

Neural encoding of vowel formant frequency in normal-hearing listeners

Physiological correlates of speech acoustics are particularly important to study in humans because it is uncertain whether animals process speech the same way humans do. Studying the physiology of speech processing in humans, however, typically requires the use of noninvasive physiological measures. This is what we attempted in a recent study (Won, Tremblay, Clinard, Wright, Sagi, and Svirsky, JASA 2016) which examined the hypothesis that neural representations of formant frequencies may help predict vowel recognition. To test the hypothesis, the frequency-following response (FFR) and vowel recognition were obtained from 38 normal-hearing listeners using four different vowels. This allowed direct comparisons between behavioral and neural data in the same individuals. FFR was used because it reflects temporal encoding of formant frequencies below about 1500 Hz. Four synthetic vowels with formant frequencies below 1500 Hz were used. Duration was 70 ms for all vowels to eliminate temporal cues and to make identification more difficult. A mathematical model (Sagi et al., JASA 2010) was used to predict vowel confusion matrices based on the neural responses. The mathematical model was successful in predicting good vs poor vowel identification performers based exclusively on physiological data.

Vowel synthesis related to equal-amplitude harmonic series in frequency ranges &amp;gt; 1 kHz combined with single harmonics &amp;lt; 1 kHz, and including variation of fundamental frequency

Front vowels can be synthesized on the basis of series of harmonics equal in amplitude, with frequencies only above 1 kHz. In these cases, spectral energy usually attributed to the first formant frequency is lacking. The present paper reports results of an experiment in which sound synthesis was performed on the basis of harmonic series covering higher frequency ranges above 1 kHz, combined with a single lower harmonic &amp;lt; 1 kHz, all harmonics equal in amplitude. Thereby, two or three sounds were synthesized for which the higher frequency range and the frequency of the lower harmonic is identical, but the frequency distance of the higher harmonics differs resulting in different perceived pitches of the sounds. Vowel recognition of all sounds was investigated by means of a listening test in which five phonetic expert listeners were asked to assign the synthesized sounds to Standard German vowel qualities. The results of the experiment reveal that the perceived vowel quality of such types of sound pairs or sound triples differs, confirming earlier indications of the spectral envelope being ambiguous with regard to vowel quality. Implications for the acoustics and perception of vowels are discussed.

Acoustic Context Alters Vowel Categorization in Perception of Noise-Vocoded Speech.

Normal-hearing listeners' speech perception is widely influenced by spectral contrast effects (SCEs), where perception of a given sound is biased away from stable spectral properties of preceding sounds. Despite this influence, it is not clear how these contrast effects affect speech perception for cochlear implant (CI) users whose spectral resolution is notoriously poor. This knowledge is important for understanding how CIs might better encode key spectral properties of the listening environment. Here, SCEs were measured in normal-hearing listeners using noise-vocoded speech to simulate poor spectral resolution. Listeners heard a noise-vocoded sentence where low-F1 (100-400Hz) or high-F1 (550-850Hz) frequency regions were amplified to encourage "eh" (/ɛ/) or "ih" (/ɪ/) responses to the following target vowel, respectively. This was done by filtering with +20dB (experiment 1a) or +5dB gain (experiment 1b) or filtering using 100% of the difference between spectral envelopes of /ɛ/ and /ɪ/ endpoint vowels (experiment 2a) or only 25% of this difference (experiment 2b). SCEs influenced identification of noise-vocoded vowels in each experiment at every level of spectral resolution. In every case but one, SCE magnitudes exceeded those reported for full-spectrum speech, particularly when spectral peaks in the preceding sentence were large (+20dB gain, 100% of the spectral envelope difference). Even when spectral resolution was insufficient for accurate vowel recognition, SCEs were still evident. Results are suggestive of SCEs influencing CI users' speech perception as well, encouraging further investigation of CI users' sensitivity to acoustic context.

Mouth Shape Recognition by Embedded Optical Sensors in Head-Mounted Display by Automated Labeling Method of Training Data Using Vowel Recognition

Mouth Shape Recognition by Embedded Optical Sensors in Head-Mounted Display by Automated Labeling Method of Training Data Using Vowel Recognition

Effect of the Number of Maxima and Stimulation Rate on Phoneme Perception Patterns Using Cochlear Implant Simulation

Purpose Maximizing speech perception for cochlear implant (CI) users can be achieved by adjusting mapping parameters. The object of this study was to investigate optimal sets of parameters of stimulation rate and the number of maxima in the CI system. Methods Listeners’ consonant and vowel perception was measured for different combinations of the number of maxima and stimulation rate using cochlear implant simulated stimuli. Twelve sets of speech stimuli were systematically created by changing the number of maxima and stimulus rate and were presented to 18 listeners with normal hearing. Results The group mean percent correct scores indicated only two pairs of parameter combinations showed significantly different results. A rate of 1,800 pps and 6 maxima resulted in significantly better consonant performance compared to a rate of 500 pps and 20 maxima. In addition, the 900 pps/8 maxima condition was significantly better compared to 500 pps/20 maxima for the vowel test. Analysis of listeners’ confusion patterns revealed they were more likely to make perception errors for the consonants /ð/, /l/, and /r/ and for the vowels /∧/, /e/, /æ/, and /ɛ/. Information transmission analysis indicated that, among other features, the voicing feature was transmitted best for consonant recognition and the backness feature was the most transmitted for vowel recognition. Conclusions The results of this study using vocoded speech with listeners with normal hearing contribute to a better understanding of CI users’ confusion patterns and possible ways to optimize cochlear implant signal processing strategies. Keywords: Cochlear implants (CI); Information transfer (IT); Speech confusion matrix; Vocoded speech; Maxima; Stimulation rate

FORMANT ANALYSIS OF BANGLA VOWEL FOR AUTOMATIC SPEECH RECOGNITION

To provide new technological benefits to the mass people, nowadays, regional and local language recognition draws attention to the researchers. Similarly to other languages, Bangla speech recognition scheme is demandable. A formant is considered as the resonance frequency of vocal tract. Formant frequencies play an important role for the purpose of automatic speech recognition, due to its noise robust characteristics. In this paper, Bangla vowels are investigated to acquire formant frequencies and its corresponding bandwidth from continuous Bangla sentences, which are considered as potential parameters for wide voice applications. For the purpose of formant analysis, cepstrum based formant estimation and Linear Predictive Coding (LPC) techniques are used. In order to acquire formant characteristics, enrich continuous sentences and widely available Bangla language corpus namely “SHRUTI” is considered. Intensive experimentation is carried out to determine formant characteristics (frequency and bandwidth) of Bangla vowels for both male and female speakers. Finally, vowel recognition accuracy of Bangla language is reported considering first three formants..