Correct Word Identification Research Articles

Acoustic-phonetic properties of Siri- and human-directed speech

Millions of people engage in spoken interactions with voice activated artificially intelligent (voice-AI) systems in their everyday lives. This study explores whether speakers have a voice-AI-specific register, relative to their speech toward an adult human. Furthermore, this study tests if speakers have targeted error correction strategies for voice-AI and human interlocutors. In a pseudo-interactive task with pre-recorded Siri and human voices, participants produced target words in sentences. In each turn, following an initial production and feedback from the interlocutor, participants repeated the sentence in one of three response types: after correct word identification, a coda error, or a vowel error made by the interlocutor. Across two studies, the rate of comprehension errors made by both interlocutors was varied (lower vs. higher error rate). Register differences are found: participants speak louder, with a lower mean f0, and with a smaller f0 range in Siri-DS. Many differences in Siri-DS emerged as dynamic adjustments over the course of the interaction. Additionally, error rate shapes how register differences are realized. One targeted error correction was observed: speakers produce more vowel hyperarticulation in coda repairs in Siri-DS. Taken together, these findings contribute to our understanding of speech register and the dynamic nature of talker-interlocutor interactions.

Intraoral voice recording-towards anew smartphone-based method for vocal rehabilitation. German version

After laryngectomy, anew voice is needed. We present the first steps in the development of asmartphone-based method. Amicrophone is placed in the mouth to record the pseudo-whispering voice of laryngectomized patients. This recording is analyzed by voice recognition software followed by voice synthesis. Eventually, this will be performed on asmartphone. We placed amicrophone at 10 different places inside and outside the mouth (two in front of the mouth (at 2 and 20 cm), five on the palate and three on the lower jaw) and made voice recordings in eight healthy men. These recordings were analyzed by voice recognition software. The text generated by the software was compared with the original text. Over all positions, the correct detection of words recorded in the mouth was 19.3% vs. 75.2% (p = 0.01) outside the mouth. In the mouth, recording taken on the maxilla (22.8%) was much better than on the mandible (13.5%) (p = 0.01). The optimum position for amicrophone on the maxilla was at the highest point of the palate with 31.9% correct word identification there (p = 0.028). Further investigations have to be undertaken with forthcoming development of smartphone processing power and with development of asmartphone-based voice recognition application.

Intraoral voice recording-towards a new smartphone-based method for vocal rehabilitation.

After laryngectomy, anew voice is needed. We present the first steps in the development of asmartphone-based method. Amicrophone is placed in the mouth to record the pseudo-whispering voice of laryngectomized patients. This recording is analyzed by voice recognition software followed by voice synthesis. Eventually, this will be performed on asmartphone. We placed amicrophone at 10 different places inside and outside the mouth (two in front of the mouth (at 2 and 20 cm), five on the palate and three on the lower jaw) and made voice recordings in eight healthy men. These recordings were analyzed by voice recognition software. The text generated by the software was compared with the original text. Over all positions, the correct detection of words recorded in the mouth was 19.3% vs. 75.2% (p = 0.01) outside the mouth. In the mouth, recording taken on the maxilla (22.8%) was much better than on the mandible (13.5%) (p = 0.01). The optimum position for amicrophone on the maxilla was at the highest point of the palate with 31.9% correct word identification there (p = 0.028). Further investigations have to be undertaken with forthcoming development of smartphone processing power and with development of asmartphone-based voice recognition application.

Spatial release from masking based on binaural processing for up to six maskers.

Spatial Release from Masking (SRM) was measured for identification of a female target word spoken in the presence of male masker words. Target words from a single loudspeaker located at midline were presented when two, four, or six masker words were presented either from the same source as the target or from spatially separated masker sources. All masker words were presented from loudspeakers located symmetrically around the centered target source in the front azimuth hemifield. Three masking conditions were employed: speech-in-speech masking (involving both informational and energetic masking), speech-in-noise masking (involving energetic masking), and filtered speech-in-filtered speech masking (involving informational masking). Psychophysical results were summarized as three-point psychometric functions relating proportion of correct word identification to target-to-masker ratio (in decibels) for both the co-located and spatially separated target and masker sources cases. SRM was then calculated by comparing the slopes and intercepts of these functions. SRM decreased as the number of symmetrically placed masker sources increased from two to six. This decrease was independent of the type of masking, with almost no SRM measured for six masker sources. These results suggest that when SRM is dependent primarily on binaural processing, SRM is effectively limited to fewer than six sound sources.

Assessment of Functional Hearing in Greek-Speaking Children Diagnosed with Central Auditory Processing Disorder.

Including speech recognition in noise testing in audiological evaluations may reveal functional hearing deficits that may otherwise remain undetected. The current study explored the potential utility of the Speech-in-Babble (SinB) test in the assessment of central auditory processing disorder (CAPD) in young children for whom diagnosis is challenging. A cross-sectional analysis. Forty-one Greek children 4-13 yr of age diagnosed with CAPD and exhibiting listening and academic problems (clinical group) and 20 age-matched controls with no listening or academic problems participated in the study. All participants' auditory processing was assessed using the same tests and instrumentation in a sound-treated room. Two equivalent lists of the SinB test, developed at the Psychoacoustic Laboratory of the Aristotle University of Thessaloniki, were administered monaurally in a counterbalanced order. SinB consists of lists of 50 phonetically balanced disyllabic words presented in background multitalker babble. Five signal-to-noise ratios (SNRs) were used in a fixed order. The children were instructed to repeat the word after each presentation. The SNR at which the child achieved 50% correct word identification served as the dependent variable or outcome measure, with higher SinB scores (measured in SNR dB) corresponding to poorer performance. SinB performance was better (lower SNR) for the normal control group versus the clinical group [F(1,35) = 43.03, p < 0.0001]. SinB inversely correlated with age for both CAPD and control groups (r = -0.648, p << 0.001 and r = -0.658, p < 0.005, respectively). Regression analysis revealed that linear models better explained the variance in the data than a quadratic model for both the control and CAPD groups. The slope (beta value of the linear model) was steeper for the clinical group compared to the control group (beta = -0.306 versus beta = -0.130, respectively). An analysis of covariance run with age as the covariate to assess the potential effect of comorbidity on SinB performance in children with CAPD with and without comorbid conditions revealed no significant differences between groups [F(1,38) = 0.149, p > 0.05]. This study offers the first detailed presentation of the performance of Greek children on a Greek language SinB test. The main finding is that SinB scores improved as a function of age in a constant manner as represented by the slope of the linear regression line for both CAPD and control groups. Results suggest that this speech recognition in competition test holds promise for differentiating typically developing Greek children from those children with CAPD across the age range studied here (4-13 yr). The SinB seemed rather immune to the presence of comorbid conditions presented by some of the children in this study, suggesting its potential utility as a valid measure of central auditory processing. While there are many speech-in-noise or competition tests in English, there are fewer in other languages. Tests like the SinB should be developed in other languages to ensure that children demonstrating "listening" problems can be properly evaluated.

Role of syllable segmentation processes in peripheral word recognition.

Previous studies of foveal visual word recognition provide evidence for a low-level syllable decomposition mechanism occurring during the recognition of a word. We investigated if such a decomposition mechanism also exists in peripheral word recognition. Single words were visually presented to subjects in the peripheral field using a 6° square gaze-contingent simulated central scotoma. In the first experiment, words were either unicolor or had their adjacent syllables segmented with two different colors (color/syllable congruent condition). Reaction times for correct word identification were measured for the two different conditions and for two different print sizes. Results show a significant decrease in reaction time for the color/syllable congruent condition compared with the unicolor condition. A second experiment suggests that this effect is specific to syllable decomposition and results from strategic, presumably involving attentional factors, rather than stimulus-driven control.

The effect of enhancing temporal periodicity cues on Cantonese tone recognition by cochlear implantees

Objectives: This study investigates the efficacy of a cochlear implant (CI) processing method that enhances temporal periodicity cues of speech. Design: Subjects participated in word and tone identification tasks. Two processing conditions – the conventional advanced combination encoder (ACE) and tone-enhanced ACE were tested. Test materials were Cantonese disyllabic words recorded from one male and one female speaker. Speech-shaped noise was added to clean speech. The fundamental frequency information for periodicity enhancement was extracted from the clean speech. Electrical stimuli generated from the noisy speech with and without periodicity enhancement were presented via direct stimulation using a Laura 34 research processor. Subjects were asked to identify the presented word. Study sample: Seven post-lingually deafened native Cantonese-speaking CI users. Results: Percent correct word, segmental structure, and tone identification scores were calculated. While word and segmental structure identification accuracy remained similar between the two processing conditions, tone identification in noise was better with tone-enhanced ACE than with conventional ACE. Significant improvement on tone perception was found only for the female voice. Conclusions: Temporal periodicity cues are important to tone perception in noise. Pitch and tone perception by CI users could be improved when listeners received enhanced temporal periodicity cues.

The effect of segmental makeup on Mandarin word and tone recognition

The present study adopts the gating paradigm to investigate the effect of onset sonorancy, and coda on Mandarin spoken word and tone recognition. Eight tonal quadruplets (all monosyllabic) with different initial segment (obstruent or sonorant), coda composition (no coda, alveolar nasal, or velar nasal) were used as the stimuli. The gates were formed by a 40-ms increment from the beginning of each word. Twenty native listeners from Taiwan were asked to listen to each stimulus, click the corresponding tone number using a mouse, write down the word and then give their confidence rating on the answer. The Isolation point (IP) based on correct word identification (both segment and tone) and the tone isolation point (TIP) were analyzed. The results indicated that tone recognition can be done before the offset of the stimuli. Tone 1 has an earlier IP than Tone 4, followed by Tones 3 and then Tone 2. Sonorant-initial words have a significant earlier IP than obstruent-initial ones. Words without coda have an earlier IP than alveolar nasal, followed by velar nasals. The processing time course for tone and segment and the effect of tonal features on word processing will be discussed.

Sensitivity to lateral information on a perceptual word identification task in french third and fifth graders

This study aimed at examining sensitivity to lateral linguistic and nonlinguistic information in third and fifth grade readers. A word identification task with a threshold was used, and targets were displayed foveally with or without distractors. Sensitivity to lateral information was inferred from the deterioration of the rate of correct word identification when displayed with distractors. Results show that the two reader groups were sensitive to both right and left lateral information. The area of sensitivity to this information was more extended for the identification of easy words than difficult words. Examination of the detrimental effect of distractors suggests that in both third and fifth graders, the impact of lateral information on foveal processing is the result of a general distraction effect, but also of linguistic processing whose nature remains to be clarified.

Sentence recognition in noise for English-, Chinese-, and Korean-native listeners.

The present study aims to investigate the effects of background noise on English sentence recognition for English-native (EN), Chinese-native (CN), and Korean-native (KN) listeners. The hearing in Noise Test (HINT) [Nilsson et al., (1994)] was used to measure the percent correct word identification in both quiet and noise conditions. Each sentence consists of uniform length with six syllables spoken by a female talker with general unaccented dialect-free American English. Two types of noise, multitalker babble and long-term speech shaped noise (LTSSN), were presented at various singal-to-noise ratios SNRs. Preliminary data demonstrated that first, the sentence recognition of non-native listeners was significantly lower than native listeners in both quiet and noise conditions; second, both native and non-native listeners showed significant amount of masking release at each SNR (−10, −5, and 0 dB SNRs); third, there was substantial individual variability in sentence recognition within each non-native group. This variability might be related with high variability in phoneme recognition in noise for the non-native individuals.

The roles of tone and syllable structure in Mandarin spoken word recognition.

The present study adopts the gating paradigm to investigate the roles of tone, onset sonorancy, and nasal coda in Mandarin spoken word recognition. Duration‐blocked gates generated from eight monosyllabic quadruplets with matching frequencies of occurrence were used as stimuli. The initial consonant of each syllable formed the first gate, with later gates formed by 40 ms increments. Twenty‐eight native Mandarin speakers from Beijing were asked to identify each gated stimulus by writing down the Chinese characters. Isolation point (IP) based on correct tone identification as well as overall correct word identification (correct onset, rhyme, and tone) were collected. Results from both conditions showed that tone 1 has an earlier IP than tone 4, which has an earlier IP than tones 2 and 3. Sonorant‐initial syllables have an earlier IP than obstruent‐initial syllables, but further analyses of covariance indicated that this is due to the fact that IP covariates with the duration of the initial consonant. Syllables without a nasal coda have an earlier IP than syllables with a nasal coda. This effect might be due to the interference of nasalization on tone perception or the delayed tonal contour realization due to the nasal coda [Xu, (1998)].

Aging, hearing acuity, and the attentional costs of effortful listening.

A dual-task interference paradigm was used to investigate the effect of perceptual effort on recall of spoken word lists by young and older adults with good hearing and with mild-to-moderate hearing loss. In addition to poorer recall accuracy, listeners with hearing loss, especially older adults, showed larger secondary task costs while recalling the word lists even though the stimuli were presented at a sound intensity that allowed correct word identification. Findings support the hypothesis that extra effort at the sensory-perceptual level attendant to hearing loss has negative consequences to downstream recall, an effect that may be further magnified with increased age.

Effect of age, presentation method, and learning on identification of noise-vocoded words

Noise vocoding was used to investigate the ability of younger and older adults with normal audiometric thresholds in the speech range to use amplitude envelope cues to identify words. In Experiment 1, four 50-word lists were tested, with each word presented initially with one frequency band and the number of bands being incremented until it was correctly identified by the listener. Both age groups required an average of 5.25 bands for 50% correct word identification and performance improved across the four lists. In Experiment 2, the same participants who completed Experiment 1 identified words in four blocked noise-vocoded conditions (16, 8, 4, 2 bands). Compared to Experiment 1, both age groups required more bands to reach the 50% correct word identification threshold in Experiment 2, 6.13, and 8.55 bands, respectively, with younger adults outperforming older adults. Experiment 3 was identical to Experiment 2 except the participants had no prior experience with noise-vocoded speech. Again, younger adults outperformed older adults, with thresholds of 6.67 and 8.97 bands, respectively. The finding of age effects in Experiments 2 and 3, but not in Experiment 1, seems more likely to be related to differences in the presentation methods than to experience with noise vocoding.

Use of supportive context by younger and older adult listeners: Balancing bottom-up and top-down information processing

Older adults often have more difficulty listening in challenging environments than their younger adult counterparts. On the one hand, auditory aging can exacerbate and/or masquerade as cognitive difficulties when auditory processing is stressed in challenging listening situations. On the other hand, an older listener can overcome some auditory processing difficulties by deploying compensatory cognitive processing, especially when there is supportive context. Supportive context may be provided by redundant cues in the external signal(s) and/or by internally stored knowledge about structures that are functionally significant in communication. It seems that listeners may achieve correct word identification in various ways depending on the challenges and supports available in complex auditory scenes. We will review evidence suggesting that older adults benefit as much or more than younger adults from supportive context at multiple levels where expectations or constraints may be related to redundancies in semantic, syntactic, lexical, phonological, or other sub-phonemic cues in the signal, and/or to expert knowledge of structures at these levels.

The effect of reduced vowel working space on speech intelligibility in Mandarin-speaking young adults with cerebral palsy

The purpose of this study was to examine the effect of reduced vowel working space on dysarthric talkers' speech intelligibility using both acoustic and perceptual approaches. In experiment 1, the acoustic-perceptual relationship between vowel working space area and speech intelligibility was examined in Mandarin-speaking young adults with cerebral palsy. Subjects read aloud 18 bisyllabic words containing the vowels /i/, /a/, and /u/ using their normal speaking rate. Each talker's words were identified by three normal listeners. The percentage of correct vowel and word identification were calculated as vowel intelligibility and word intelligibility, respectively. Results revealed that talkers with cerebral palsy exhibited smaller vowel working space areas compared to ten age-matched controls. The vowel working space area was significantly correlated with vowel intelligibility (r=0.632, p<0.005) and with word intelligibility (r=0.684, p<0.005). Experiment 2 examined whether tokens of expanded vowel working spaces were perceived as better vowel exemplars and represented with greater perceptual spaces than tokens of reduced vowel working spaces. The results of the perceptual experiment support this prediction. The distorted vowels of talkers with cerebral palsy compose a smaller acoustic space that results in shrunken intervowel perceptual distances for listeners.

A scale for hearing aid benefit measures and auditory disability.

Quoting disability scores or measures of hearing aid benefit based on percent correct word identification or on dB signal-to-noise ratio at criterion performance poses a communication problem as scientists and clinicians may have no intuitive understanding of the scales involved. In particular, percent correct does not allow direct comparison of differences obtained at different performance levels, for example by different subject groups. In circumstances where the better ear hearing levels are known to be a major determinant of disability, it is both possible and legitimate to transform performance measures such as percent correct into a dB HL equivalent. In two sets of data from clinical research projects this relationship is shown to be linear, allowing a particularly simple transformation and giving an already understood scale. This transformation procedure has been cross-validated against the acoustic gain used in a sample of patients with conductive losses where the hearing loss is purely an attenuation. The procedure is offered as a metric for expressing the benefits of management.

Parafoveal word perception: A further case against semantic preprocessing.

Three experiments were performed to explore the processing of parafoveal information. A foveal and a parafoveal word pair were presented within the time frame of a single fixation, but short enough to prevent successful refixation of the parafoveal word. Subjects reporting partial parafoveal word information were not able to use this information to bias their choice in a forced-choice task requiring some semantic information from the parafoveal word for correct performance. This result was observed even though correct parafoveal word identification was facilitated if the foveal stimulus was related to the parafoveal word. Further, the data strongly suggest that information is obtained initially from the fixated words, regardless of whether foveal word presentation precedes or follows the parafoveal stimulus.

Differences in the sequential integration of form as a function of age and interstimulus interval

The sequential integration of visual stimuli was studied in 12 young (18-28 years) and 12 old (51-62 years) men and women. Stimuli were constructed as corresponding word halves and inter-stimulus intervals (ISI's) of varied duration (0-150 msec.) were introduced between the presentation of the two halves. Correct word identification was found to be an inverse function of ISI for both young and old subjects. The number of stimuli correctly identified was significantly lower for the older subjects at all levels of ISI. None of the interactions of age, sex, and ISI was significant. The data appear to question the "stimulus persistence" model as applied to age differences in the temporal organization of form.

Temporal anagrams: Word identification with successively presented letters in scrambled order

High-frequency seven-letter words were displayed letter by letter in scrambled temporal orders upon an oscilloscope. After each incorrect identification, the sequence was repeated. Highest word-identification scores were obtained with initial and terminal starting positions, with shortest and longest spacingbetween successive letters, and at the slowest and fastest rates of letter presentation. The relative contribution of each variable was maintained in combination with other variables. The conditional probability of a correct word identification, given that the word was not identified on a previous presentation, plunges sharply with successive presentations. This apparent nonindependence of performance over successive presentations is, however, shown to be an artifact of pooling the results of heterogeneous experimental variables.