Word Recognition Accuracy Research Articles

Limitations of episodic memory for highly similar auditory stimuli

ABSTRACTTwo experiments investigated participants’ recognition memory for word content, while varying vocal characteristics, and for vocal characteristics alone. In Experiment 1, participants performed an auditory recognition task in which they identified whether a spoken word was “new”, “old” (repeated word, repeated voice), or “similar” (repeated word, new voice). Results showed that word recognition accuracy was lower for similar trials than old trials. In Experiment 2, participants performed an auditory recognition task in which they identified whether or not a phrase was spoken in an old or new voice, with repetitions occurring after a variable number of intervening stimuli. Results showed that recognition accuracy was lower when old voices spoke an alternate message than a repeated message and accuracy decreased as a function of number of intervening items. Overall, the results suggest that speech recognition is better for lexical content than vocal characteristics alone.

Effects of Hearing Loss on Dual-Task Performance in an Audiovisual Virtual Reality Simulation of Listening While Walking.

Most activities of daily living require the dynamic integration of sights, sounds, and movements as people navigate complex environments. Nevertheless, little is known about the effects of hearing loss (HL) or hearing aid (HA) use on listening during multitasking challenges. The objective of the current study was to investigate the effect of age-related hearing loss (ARHL) on word recognition accuracy in a dual-task experiment. Virtual reality (VR) technologies in a specialized laboratory (Challenging Environment Assessment Laboratory) were used to produce a controlled and safe simulated environment for listening while walking. In a simulation of a downtown street intersection, participants completed two single-task conditions, listening-only (standing stationary) and walking-only (walking on a treadmill to cross the simulated intersection with no speech presented), and a dual-task condition (listening while walking). For the listening task, they were required to recognize words spoken by a target talker when there was a competing talker. For some blocks of trials, the target talker was always located at 0° azimuth (100% probability condition); for other blocks, the target talker was more likely (60% of trials) to be located at the center (0° azimuth) and less likely (40% of trials) to be located at the left (270° azimuth). The participants were eight older adults with bilateral HL (mean age = 73.3 yr, standard deviation [SD] = 8.4; three males) who wore their own HAs during testing and eight controls with normal hearing (NH) thresholds (mean age = 69.9 yr, SD = 5.4; two males). No participant had clinically significant visual, cognitive, or mobility impairments. Word recognition accuracy and kinematic parameters (head and trunk angles, step width and length, stride time, cadence) were analyzed using mixed factorial analysis of variances with group as a between-subjects factor. Task condition (single versus dual) and probability (100% versus 60%) were within-subject factors. In analyses of the 60% listening condition, spatial expectation (likely versus unlikely) was a within-subject factor. Differences between groups in age and baseline measures of hearing, mobility, and cognition were tested using t tests. The NH group had significantly better word recognition accuracy than the HL group. Both groups performed better when the probability was higher and the target location more likely. For word recognition, dual-task costs for the HL group did not depend on condition, whereas the NH group demonstrated a surprising dual-task benefit in conditions with lower probability or spatial expectation. For the kinematic parameters, both groups demonstrated a more upright and less variable head position and more variable trunk position during dual-task conditions compared to the walking-only condition, suggesting that safe walking was prioritized. The HL group demonstrated more overall stride time variability than the NH group. This study provides new knowledge about the effects of ARHL, HA use, and aging on word recognition when individuals also perform a mobility-related task that is typically experienced in everyday life. This research may help inform the development of more effective function-based approaches to assessment and intervention for people who are hard-of-hearing.

Spoken word recognition in children with autism spectrum disorder: The role of visual disengagement.

Deficits in visual disengagement are one of the earliest emerging differences in infants who are later diagnosed with autism spectrum disorder. Although researchers have speculated that deficits in visual disengagement could have negative effects on the development of children with autism spectrum disorder, we do not know which skills are disrupted or how this disruption takes place. As a first step in understanding this issue, this study investigated the relationship between visual disengagement and a critical skill in early language development: spoken word recognition. Participants were 18 children with autism spectrum disorder (aged 4-7 years). Consistent with our predictions, children with poorer visual disengagement were slower and less accurate to process familiar words; disengagement explained over half of the variance in spoken word recognition. Visual disengagement remained uniquely associated with spoken word recognition after accounting for children's vocabulary size and age. These findings align with a recently proposed developmental model in which poor visual disengagement decreases the speed and accuracy of real-time spoken word recognition in children with autism spectrum disorder-which, in turn, may negatively affect their language development.

Reading Fluency and College Readiness

AbstractThe Common Core State Standards suggest that an appropriate goal for secondary education is college and career readiness. Previous research has identified reading fluency as a critical component for proficient reading. One component of fluency is word recognition accuracy and automaticity. The present study attempted to determine the word recognition accuracy and automaticity indicators for incoming college students and to examine the relationship between oral reading rate and ACT scores. Knowledge of such performance may provide secondary literacy educators with another tool for assessing reading proficiency and college readiness. Eighty‐one students were asked to read a college‐level narrative passage. Mean student accuracy and automacity scores were determined, as well as correlations between accuracy, automacity, and students’ ACT scores. Results suggest a moderate and significant relationship between measures of automaticity and both the ACT reading subtest and ACT composite scores.

The Bilingual Disadvantage in Speech Understanding in Noise Is Likely a Frequency Effect Related to Reduced Language Exposure.

The present study sought to explain why bilingual speakers are disadvantaged relative to monolingual speakers when it comes to speech understanding in noise. Exemplar models of the mental lexicon hold that each encounter with a word leaves a memory trace in long-term memory. Words that we encounter frequently will be associated with richer phonetic representations in memory and therefore recognized faster and more accurately than less frequently encountered words. Because bilinguals are exposed to each of their languages less often than monolinguals by virtue of speaking two languages, they encounter all words less frequently and may therefore have poorer phonetic representations of all words compared to monolinguals. In the present study, vocabulary size was taken as an estimate for language exposure and the prediction was made that both vocabulary size and word frequency would be associated with recognition accuracy for words presented in noise. Forty-eight early Spanish–English bilingual and 53 monolingual English young adults were tested on speech understanding in noise (SUN) ability, English oral verbal ability, verbal working memory (WM), and auditory attention. Results showed that, as a group, monolinguals recognized significantly more words than bilinguals. However, this effect was attenuated by language proficiency; higher proficiency was associated with higher accuracy on the SUN test in both groups. This suggests that greater language exposure is associated with better SUN. Word frequency modulated recognition accuracy and the difference between groups was largest for low frequency words, suggesting that the bilinguals’ insufficient exposure to these words hampered recognition. The effect of WM was not significant, likely because of its large shared variance with language proficiency. The effect of auditory attention was small but significant. These results are discussed within the Ease of Language Understanding model (Rönnberg et al., 2013), which provides a framework for explaining individual differences in SUN.

Is visual content in textual search interfaces beneficial to dyslexic users?

Dyslexia is a learning disability characterised by problems with accurate or fluent word recognition, poor decoding, and poor spelling abilities. Although several studies have addressed dyslexia and Web accessibility, less is known about how dyslexia affects information search. This study investigated whether the inclusion of icons in search user interfaces enhances performance among dyslexics. A total of 21 dyslexics and 21 controls completed 52 search tasks in 4 conditions: icons only, words only, and both icons and words in a grid layout and a list layout, while eye movements were recorded. Dyslexics took significantly longer than controls to locate targets in tasks containing text, but not in the icon-only condition. Dyslexics had longer fixation durations than controls in both icon and text based search arrays, suggesting higher mental load associated with search tasks generally. The addition of words to icon arrays led to faster search times within controls, but not dyslexics. Dyslexics also exhibited more fixations on dual-modality tasks, and longer scanpaths than controls in list layout. Both groups were fastest searching the list layout, with icons and words listed in columns. Results are discussed in terms of the design of accessible search interfaces for dyslexic users, taking into account mental load of dual-modality information display, and the arrangement of search items. Empirical data is provided for the design of accessible search results interfaces for dyslexics. HighlightsDyslexics are generally slower in visual search arrays containing text.Including icons in arrays seems distracting for both dyslexics and controls.Including icons in lists seems beneficial to performance generally.Content mode and layout interact to effect search performance in dyslexia.

Modeling State-Conditional Observation Distribution Using Weighted Stereo Samples for Factorial Speech Processing Models

This paper investigates the effectiveness of factorial speech processing models in noise-robust automatic speech recognition tasks. For this purpose, the paper proposes an idealistic approach for modeling state-conditional observation distribution of factorial models based on weighted stereo samples. This approach is an extension to previous single pass retraining for ideal model compensation which is extended here to support multiple audio sources. Non-stationary noises can be considered as one of these audio sources with multiple states. Experiments of this paper over the set A of the Aurora 2 dataset show that recognition performance can be improved by this consideration. The improvement is significant in low signal to noise energy conditions, up to 4% absolute word recognition accuracy. In addition to the power of the proposed method in accurate representation of state-conditional observation distribution, it has an important advantage over previous methods by providing the opportunity to independently select feature spaces for both source and corrupted features. This opens a new window for seeking better feature spaces appropriate for noisy speech, independent from clean speech features.

Children’s perception of native dialects and nonnative accents

In quiet listening conditions, school-aged children can have difficulty understanding nonnative-accented speech whereas adults tend to be highly accurate. The addition of noise substantially depresses word recognition accuracy for both groups. Here, these findings are extended to the perception of an unfamiliar native dialect. Children between the ages of 5 and 7 years (n = 90) were presented with HINT-C sentences produced by three female talkers with different accents—American English (midland dialect), British English, and Japanese-accented English—in quiet or in 8-talker babble with a + 4 dB SNR. Results showed highly significant main effects of accent, listening condition (noise, quiet), and age in the expected directions, as well as an interaction between talker accent and listening condition. In quiet, children showed very accurate word recognition for the American and British talkers (97% and 95% correct, respectively) with lower accuracy for the nonnative talker (73% correct). Compared to the quiet condition, performance declined more in the noise-added condition for the British (20% decline) and nonnative talkers (21%) than for the American talker (7%). These results suggest that although school-aged children can understand unfamiliar native dialects, their representations of these dialects may be fragile and highly susceptible to environmental degradation.

An fMRI study investigating effects of conceptually related sentences on the perception of degraded speech

Prior research has shown that the perception of degraded speech is influenced by within sentence meaning and recruits one or more components of a frontal–temporal–parietal network. The goal of the current study is to examine whether the overall conceptual meaning of a sentence, made up of one set of words, influences the perception of a second acoustically degraded sentence, made up of a different set of words. Using functional magnetic resonance imaging (fMRI), we presented an acoustically clear sentence followed by an acoustically degraded sentence and manipulated the semantic relationship between them: Related in meaning (but consisting of different content words), Unrelated in meaning, or Same. Results showed that listeners' word recognition accuracy for the acoustically degraded sentences was significantly higher when the target sentence was preceded by a conceptually related compared to a conceptually unrelated sentence. Sensitivity to conceptual relationships was associated with enhanced activity in middle and inferior frontal, temporal, and parietal areas. In addition, the left middle frontal gyrus (LMFG), left inferior frontal gyrus (LIFG), and left middle temporal gyrus (LMTG) showed activity that correlated with individual performance on the Related condition. The superior temporal gyrus (STG) showed increased activation in the Same condition suggesting that it is sensitive to perceptual similarity rather than the integration of meaning between the sentence pairs. A fronto-temporo-parietal network appears to consolidate information sources across multiple levels of language (acoustic, lexical, syntactic, semantic) to build, and ultimately integrate conceptual information across sentences and facilitate the perception of a degraded speech signal. However, the nature of the sources of information that are available differentially recruit specific regions and modulate their activity within this network. Implications of these findings for the functional architecture of the network are considered.

Synthesis of Common Arabic Handwritings to Aid Optical Character Recognition Research

Document analysis tasks such as pattern recognition, word spotting or segmentation, require comprehensive databases for training and validation. Not only variations in writing style but also the used list of words is of importance in the case that training samples should reflect the input of a specific area of application. However, generation of training samples is expensive in the sense of manpower and time, particularly if complete text pages including complex ground truth are required. This is why there is a lack of such databases, especially for Arabic, the second most popular language. However, Arabic handwriting recognition involves different preprocessing, segmentation and recognition methods. Each requires particular ground truth or samples to enable optimal training and validation, which are often not covered by the currently available databases. To overcome this issue, we propose a system that synthesizes Arabic handwritten words and text pages and generates corresponding detailed ground truth. We use these syntheses to validate a new, segmentation based system that recognizes handwritten Arabic words. We found that a modification of an Active Shape Model based character classifiers—that we proposed earlier—improves the word recognition accuracy. Further improvements are achieved, by using a vocabulary of the 50,000 most common Arabic words for error correction.

Noise Robust Speech Recognition Using Deep Belief Networks

Most current speech recognition systems use Hidden Markov Models (HMMs) to deal with the temporal variability of speech and Gaussian mixture models (GMMs) to determine how well each state of each HMM fits a frame or a short window of frames of coefficients that represents the acoustic input. In these systems acoustic inputs are represented by Mel Frequency Cepstral Coefficients temporal spectrogram known as frames. But MFCC is not robust to noise. Consequently, with different train and test conditions the accuracy of speech recognition systems decreases. On the other hand, using MFCCs of larger window of frames in GMMs needs more computational power. In this paper, Deep Belief Networks (DBNs) are used to extract discriminative information from larger window of frames. Nonlinear transformations lead to high-order and low-dimensional features which are robust to variation of input speech. Multiple speaker isolated word recognition tasks with 100 and 200 words in clean and noisy environments has been used to test this method. The experimental results indicate that this new method of feature encoding result in much better word recognition accuracy.

F0 contour approximation model for a one-stream tonal word recognition system

The performance of a non-tonal speech recognition system degrades when confronted with the task of recognizing tonal words. Several speech recognition applications require tonal word recognition. Therefore, this paper considers how to create a suitable tone model for a tonal syllable recognition system serving application devices based on a one-stream scheme. The fundamental frequency contour (F0 contour) approximation model is proposed here to estimate F0 continuity contours for all of a tonal word. The processes of approximation include voice detection, F0 smoothing, F0 forecasting, and F0 normalization. To model the F0 contours of unvoiced regions belonging to F0 forecasting, a linear regression function is used to create an approximate F0 contour. Experimental results indicate that the proposed model improves the accuracy of tonal word recognition by 8.6% and 12.2%, respectively, compared with conventional random and exponential approaches.

Many neighborhoods: Phonological and perceptual neighborhood density in lexical production and perception

Abstract We examine the relationship of lexical representations, pronunciation variation, and word recognition, by investigating effects of two lexical variables: Phonological Neighborhood Density (the number of words that can be formed by a single phoneme substitution, addition, or deletion from the target word), as well as a measure of the perceptual similarity of a target word to other words in the lexicon. We show that perceptual similarity to other words affects recognition, but not production. Phonological Neighborhood Density, on the other hand, affects both word durations and recognition accuracy (words with many neighbors shorten and are difficult recognition targets). We interpret our results as indicating that effects of Phonological Neighborhood Density on pronunciation are not generally due to perceptual similarity of the target to other words. Our results are consistent with a more general line of research demonstrating effects of ‘central’ processes on ‘peripheral’ processes such as articulation, as well as effects of modality-specific properties, such as auditory similarity and motor movements, on measures thought to tap central processes.

How older adults use cognition in sentence-final word recognition

ABSTRACTThis study examined the effects of executive control and working memory on older adults’ sentence-final word recognition. The question we addressed was the importance of executive functions to this process and how it is modulated by the predictability of the speech material. To this end, we tested 173 neurologically intact adult native English speakers aged 55–84 years. Participants were given a sentence-final word recognition test in which sentential context was manipulated and sentences were presented in different levels of babble, and multiple tests of executive functioning assessing inhibition, shifting, and efficient access to long-term memory, as well as working memory. Using a generalized linear mixed model, we found that better inhibition was associated with higher accuracy in word recognition, while increased age and greater hearing loss were associated with poorer performance. Findings are discussed in the framework of semantic control and are interpreted as supporting a theoretical view of executive control which emphasizes functional diversity among executive components.

Fingerspelling as a Novel Gateway into Reading Fluency in Deaf Bilinguals.

Studies have shown that American Sign Language (ASL) fluency has a positive impact on deaf individuals’ English reading, but the cognitive and cross-linguistic mechanisms permitting the mapping of a visual-manual language onto a sound-based language have yet to be elucidated. Fingerspelling, which represents English orthography with 26 distinct hand configurations, is an integral part of ASL and has been suggested to provide deaf bilinguals with important cross-linguistic links between sign language and orthography. Using a hierarchical multiple regression analysis, this study examined the relationship of age of ASL exposure, ASL fluency, and fingerspelling skill on reading fluency in deaf college-age bilinguals. After controlling for ASL fluency, fingerspelling skill significantly predicted reading fluency, revealing for the first-time that fingerspelling, above and beyond ASL skills, contributes to reading fluency in deaf bilinguals. We suggest that both fingerspelling—in the visual-manual modality—and reading—in the visual-orthographic modality—are mutually facilitating because they share common underlying cognitive capacities of word decoding accuracy and automaticity of word recognition. The findings provide support for the hypothesis that the development of English reading proficiency may be facilitated through strengthening of the relationship among fingerspelling, sign language, and orthographic decoding en route to reading mastery, and may also reveal optimal approaches for reading instruction for deaf and hard of hearing children.

New interface for musical instruments using lip reading

As smart audio-visual multimedia devices are developed for various applications, there has been a growing interest in effective human–computer interaction (HCI) interfaces for specific environments. There have also been great efforts to implement HCI interfaces into musical instruments, in which it would be possible to take intuitions, comfort and expressiveness into the musical instruments. However, most of the traditional HCI interfaces are not applicable because both hands are likely to be occupied while playing a musical instrument. In this environment, a lip reading method can be used. A lip reading method is a HCI method that analyses lip motion to recognise spoken words. In this study, a lip reading method is proposed with its application. As a specific example of the interface for musical instruments, the guitar effector application is presented. The proposed lip reading method uses a constrained local model instead of the conventional active appearance model for effective facial feature tracking. The proposed method also uses a dynamic time warping-based classifier for word recognition which is effective for simple real-time implementation of lip reading. The proposed lip reading method shows 85.0% word recognition accuracy on OuluVS database and is effectively applied to the proposed guitar effector application.

The Push for More Challenging Texts: An Analysis of Early Readers’ Rate, Accuracy, and Comprehension

The purpose of the study was to examine the relationship between the challenge level of text and early readers’ reading comprehension. This relationship was also examined with consideration to students’ word recognition accuracy and reading rate. Participants included 636 students, in Grades 1–3, in a southeastern state. Results suggest that students reading texts well above their actual grade levels, even with sufficient accuracy scores, scored significantly lower on comprehension than students reading texts at their actual grade level. This result also held regardless of students’ reading rates. Findings signal the importance of considering text level during instruction and suggest some caution is warranted when pushing students into texts well above their grade levels.

EEG-based time and spatial interpretation of activation areas for relaxation and words writing between poor and capable dyslexic children.

Symptoms of dyslexia such as difficulties with accurate and/or fluent word recognition, and/or poor spelling as well as decoding abilities, are easily misinterpreted as laziness and defiance amongst school children. Indeed, 37.9% of 699 school dropouts and failures are diagnosed as dyslexic. Currently, Screening for dyslexia relies heavily on therapists, whom are few and subjective, yet objective methods are still unavailable. EEG has long been a popular method to study the cognitive processes in human such as language processing and motor activity. However, its interpretation is limited to time and frequency domain, without visual information, which is still useful. Here, our research intends to illustrate an EEG-based time and spatial interpretation of activated brain areas for the poor and capable dyslexic during the state of relaxation and words writing, being the first attempt ever reported. From the 2D distribution of EEG spectral at the activation areas and its progress with time, it is observed that capable dyslexics are able to relax compared to poor dyslexics. During the state of words writing, neural activities are found higher on the right hemisphere than the left hemisphere of the capable dyslexics, which suggests a neurobiological compensation pathway in the right hemisphere, during reading and writing, which is not observed in the poor dyslexics.

English

English

Model-based clustered sparse imputation for noise robust speech recognition

In the sparse imputation approach, missing spectral components of speech are estimated using the compressive sensing technique. For this purpose, a dictionary of clean speech components must be prepared. Noisy feature vectors are then reconstructed by the dictionary queries. In this approach, the dictionary elements should adequately cover all the possible varieties of the speech feature vectors; so, for a subject speech frame, there will be lots of irrelevant components inside the dictionary. These components make a huge size for the dictionary that in turn, slow down the estimation process and may produce artifacts in the final estimation. To face this problem, the current work proposes to cluster the dictionary queries in some smaller subspaces; the relevant subspace for a subject feature vector could be found through the posterior criterion. Moreover, this is shown that the likelihood of the Gaussian models developed for the subspaces could role as a regularization term and act as an extra prior knowledge in the estimation process and increases the final performance, significantly. To evaluate the benefits of the proposed methods, some well-designed ASR experiments are conducted on two different speech corpora, an English noisy connected digit database (Aurora 2) and a Persian continuous speech corpus (FARSDAT). The experiments show that the proposed methods not only increase the absolute word recognition accuracy but also make the entire process few times faster than the original sparse imputation approach.