High Talker Variability Research Articles

Assessment of High-variability Speech Recognition in Adult Cochlear Implant Users using PRESTO.

Speech recognition in adult cochlear implant (CI) users is typically assessed using sentence materials with low talker variability. Little is known about the effects of talker variability on speech recognition in adult CI users, the factors underlying individual differences in speech recognition with high talker variability, or how sentence materials with high talker variability could be utilized clinically. To examine the effects of talker variability on sentence recognition in adult CI users, using sentences from the Perceptually Robust English Sentence Test Open-Set (PRESTO), and to examine the relation between working memory capacity and high-variability speech recognition. Postlingually deafened adult CI users and normal-hearing (NH) listeners under CI simulation completed sentence recognition tests that contained varying levels of talker variability, including HINT (low-variability), AzBio (moderate-variability), and PRESTO sentences (high-variability). The tasks were completed in both quiet and multi-talker babble (MTB). For the adult CI users only, the relation between sentence recognition accuracy and working memory capacity was assessed. Twenty postlingually deafened adult CI users and 35 NH adults under 8-channel acoustic noise-vocoder simulations of CI hearing. In both CI and NH groups, performance decreased as a function of increased talker variability, with the best scores obtained on HINT (low-variability), then AzBio (moderate-variability), followed by PRESTO (high-variability) in quiet. In MTB, performance was significantly lower on PRESTO sentences, compared to HINT and AzBio sentences, which were not significantly different. Working memory capacity in the CI users was related to sentence recognition accuracy across all materials and conditions. Findings from the current study suggest that the increased talker variability in the PRESTO sentence materials has a detrimental effect on speech recognition in both adult CI users and NH listeners under CI simulation, particularly when speech is further degraded by MTB. For adult CI users, working memory capacity contributes to speech recognition abilities. Sentence recognition testing with high-variability, multi-talker materials, as in PRESTO, provides robust assessment of speech recognition abilities for research and clinical application, generating a wide range of scores for evaluating individual differences without ceiling effects when compared to conventional low-variability sentences.

Differential relations of auditory and cognitive functions to speech recognition tasks in adult cochlear implant users

Clinical testing for assessing adult cochlear implant (CI) candidacy and outcomes relies primarily on tasks of open-set recognition for words and sentences. However, it is not clear in what ways performance on these different tasks relates differentially to auditory and cognitive functions. We tested the hypothesis that auditory and cognitive functions would contribute differentially to performance on recognition of isolated words (CID-W22 words), meaningful sentences (Harvard Standard), nonsense sentences (Harvard Anomalous), and sentences with high talker variability (PRESTO) in adult CI users. Sixty-five experienced CI users completed a battery of auditory and cognitive tests to assess auditory spectro-temporal processing, vocabulary size, working memory capacity, inhibition-concentration, nonverbal reasoning, and speed of lexical access. Findings were that spectro-temporal processing, lexical access speed, and nonverbal reasoning predicted scores across all word and sentence recognition tests. In contrast, inhibition-concentration contributed to nonsense sentence recognition, while vocabulary knowledge contributed to high talker variability sentence recognition. Results suggest that there is a core set of auditory and cognitive functions that contribute broadly to speech recognition, while more specific cognitive functions contribute to performance on some speech recognition tasks. Clinical implications are that individual cognitive abilities should be considered when evaluating CI candidacy and outcomes.

Processing of linguistic and indexical information in adult cochlear implant users

Normal-hearing (NH) listeners simultaneously make use of both linguistic (speech content) and indexical (talker-specific) information in speech perception. For adult cochlear implant (CI) users, limitations in the talker-specific details conveyed by their devices may impact processing dependences between linguistic and indexical information. The current study examined the extent to which linguistic and indexical information are processed interdependently in adult CI users, using speeded phoneme and talker gender classification tasks. Higher-performing adult CI users (open-set word recognition >65% correct) reported the target vowel (phoneme classification) or target talker gender (gender classification). The non-target dimension varied by the number of talkers (phoneme classification) and the number of words (gender classification). Largely consistent with NH findings, adult CI users displayed asymmetric interference between linguistic and indexical dimensions. Phoneme classification was less accurate and slower with high talker variability. In contrast, gender classification was not affected by word variability. Processing dependencies also varied across individual CI users and related to auditory and cognitive-linguistic abilities. These preliminary findings suggest that adult CI users may make use of degraded linguistic and indexical information during speech processing. However, the extent to which they do so may depend on individual auditory and cognitive-linguistic factors.

Are There Real-world Benefits to Bimodal Listening?

To assess the benefits of bimodal listening (i.e., addition of contralateral hearing aid) for cochlear implant (CI) users on real-world tasks involving high-talker variability speech materials, environmental sounds, and self-reported quality of life (quality of hearing) in listeners' own best-aided conditions. Cross-sectional study between groups. Outpatient hearing clinic. Fifty experienced adult CI users divided into groups based on normal daily listening conditions (i.e., best-aided conditions): unilateral CI (CI), unilateral CI with contralateral HA (bimodal listening; CIHA), or bilateral CI (CICI). Task-specific measures of speech recognition with low (Harvard Standard Sentences) and high (Perceptually Robust English Sentence Test Open-set corpus) talker variability, environmental sound recognition (Familiar Environmental Sounds Test-Identification), and hearing-related quality of life (Nijmegen Cochlear Implant Questionnaire). Test group differences among CI, CIHA, and CICI conditions. No group effect was observed for speech recognition with low or high-talker variability, or hearing-related quality of life. Bimodal listeners demonstrated a benefit in environmental sound recognition compared with unilateral CI listeners, with a trend of greater benefit than the bilateral CI group. There was also a visual trend for benefit on high-talker variability speech recognition. Findings provide evidence that bimodal listeners demonstrate stronger environmental sound recognition compared with unilateral CI listeners, and support the idea that there are additional advantages to bimodal listening after implantation other than speech recognition measures, which are at risk of being lost if considering bilateral implantation.

High- and Low-Performing Adult Cochlear Implant Users on High-Variability Sentence Recognition: Differences in Auditory Spectral Resolution and Neurocognitive Functioning.

Postlingually deafened adult cochlear implant (CI) users routinely display large individual differences in the ability to recognize and understand speech, especially in adverse listening conditions. Although individual differences have been linked to several sensory (''bottom-up'') and cognitive (''top-down'') factors, little is currently known about the relative contributions of these factors in high- and low-performing CI users. The aim of the study was to investigate differences in sensory functioning and neurocognitive functioning between high- and low-performing CI users on the Perceptually Robust English Sentence Test Open-set (PRESTO), a high-variability sentence recognition test containing sentence materials produced by multiple male and female talkers with diverse regional accents. CI users with accuracy scores in the upper (HiPRESTO) or lower quartiles (LoPRESTO) on PRESTO in quiet completed a battery of behavioral tasks designed to assess spectral resolution and neurocognitive functioning. Twenty-one postlingually deafened adult CI users, with 11 HiPRESTO and 10 LoPRESTO participants. A discriminant analysis was carried out to determine the extent to which measures of spectral resolution and neurocognitive functioning discriminate HiPRESTO and LoPRESTO CI users. Auditory spectral resolution was measured using the Spectral-Temporally Modulated Ripple Test (SMRT). Neurocognitive functioning was assessed with visual measures of working memory (digit span), inhibitory control (Stroop), speed of lexical/phonological access (Test of Word Reading Efficiency), and nonverbal reasoning (Raven's Progressive Matrices). HiPRESTO and LoPRESTO CI users were discriminated primarily by performance on the SMRT and secondarily by the Raven's test. No other neurocognitive measures contributed substantially to the discriminant function. High- and low-performing CI users differed by spectral resolution and, to a lesser extent, nonverbal reasoning. These findings suggest that the extreme groups are determined by global factors of richness of sensory information and domain-general, nonverbal intelligence, rather than specific neurocognitive processing operations related to speech perception and spoken word recognition. Thus, although both bottom-up and top-down information contribute to speech recognition performance, low-performing CI users may not be sufficiently able to rely on neurocognitive skills specific to speech recognition to enhance processing of spectrally degraded input in adverse conditions involving high talker variability.

The role of talker variability in the perceptual learning of Mandarin tones by American English listeners

Abstract Research on segmentals has suggested that a key component of High Variability Phonetic Training (HVPT) is high talker variability. However, the extent to which high talker variability improves perception of tones is unclear. This study examined the effects of high talker variability on the perception of Mandarin tones (Tones 1–4) by English-speaking listeners. A training paradigm that used multiple talkers (multitalker group) was compared with a paradigm that used one talker (single-talker group). The results showed that the multitalker group outperformed the single-talker group, and they retained their learning better than the single-talker group did for 6 months. Neither group, however, improved their perception of Tone 1 or generalized their learning of monosyllables to disyllables. The results suggest that although high talker variability can effectively improve tone perception, it does not improve the perception of more tone categories or yield generalization of learning to more contexts compared to low talker variability.

Training-induced brain activation and functional connectivity differentiate multi-talker and single-talker speech training.

In second language acquisition studies, the high talker variability training approach has been frequently used to train participants to learn new speech patterns. However, the neuroplasticity induced by training is poorly understood. In the present study, native English speakers were trained on non-native pitch patterns (linguistic tones from Mandarin Chinese) in multi-talker (N = 16) or single-talker (N = 16) training conditions. We focused on two aspects of multi-talker training, voice processing and lexical phonology accessing, and used functional magnetic resonance imaging (fMRI) to measure the brain activation and functional connectivity (FC) of two regions of interest in a tone identification task conducted before and after training, namely the anterior part of the right superior temporal gyrus (aRSTG) and the posterior left superior temporal gyrus (pLSTG). The results showed distinct patterns of associations between neural signals and learning success for multi-talker training. Specifically, post-training brain activation in the aRSTG and FC strength between the aRSTG and pLSTG were correlated with learning success in the multi-talker training group but not in the single-talker group. These results suggest that talker variability in the training procedure may enhance neural efficiency in these brain areas and strengthen the cooperation between them. Our findings highlight the brain processing of newly learned speech patterns is influenced by the given training approach.