Multiple Talkers Research Articles

Brain-Controlled Augmented Hearing for Spatially Moving Conversations in Multi-Talker Environments.

Focusing on a specific conversation amidst multiple interfering talkers is challenging, especially for those with hearing loss. Brain-controlled assistive hearing devices aim to alleviate this problem by enhancing the attended speech based on the listener's neural signals using auditory attention decoding (AAD). Departing from conventional AAD studies that relied on oversimplified scenarios with stationary talkers, a realistic AAD task that involves multiple talkers taking turns as they continuously move in space in background noise is presented. Invasive electroencephalography (iEEG) data are collected from three neurosurgical patients as they focused on one of the two moving conversations. An enhanced brain-controlled assistive hearing system that combines AAD and a binaural speaker-independent speech separation model is presented. The separation model unmixes talkers while preserving their spatial location and provides talker trajectories to the neural decoder to improve AAD accuracy. Subjective and objective evaluations show that the proposed system enhances speech intelligibility and facilitates conversation tracking while maintaining spatial cues and voice quality in challenging acoustic environments. This research demonstrates the potential of this approach in real-world scenarios and marks a significant step toward developing assistive hearing technologies that adapt to the intricate dynamics of everyday auditory experiences.

Hearing at the Mall: Multibeam Processing Technology Improves Hearing Group Conversations in a Real-World Food Court Environment.

This study aimed to evaluate conversation hearing with an adaptive beamforming hearing aid that supports adaptive tracking of multiple talkers in an ecologically valid, real-world food court environment in a busy mall. Twenty older adult experienced hearing aid wearers with sensorineural hearing loss were fitted in the lab with binaural receiver-in-the-canal style hearing aids set with two programs, each having a different beamforming strategy. The participant and two researchers then met in a moderately noisy and reverberant food court at a local mall where the participant was asked to listen to a conversation between the two researchers. Participants rated the extent of their agreement with 10 positively worded statements specific to the conversation twice, once for each program. Participants then provided program-preference ratings for seven different aspects of a conversation during which the programs were switched back and forth by the researcher, so that participants were unaware of the condition to which they were listening. Real-world subjective ratings for all domains resulted in positive values on average for both programs. Pairwise comparisons indicated that the intervention algorithm had higher absolute ratings on five of the 10 criteria including understanding, clarity, focus, listening effort, and background noise. Ratings for preference between programs indicated a significant preference for the intervention algorithm for all seven criteria. In a real-world setting, the use of hearing aids with separate processing of sounds from the front and back hemisphere provided positive subjective ratings. However, following a group conversation with multiple conversation partners, improvements in the algorithm to account for the locations of interlocutors and the natural head turning of the hearing aid wearer that occurs during a conversation by adding and controlling multiple adaptive beams in the front hemisphere significantly influenced preference for all aspects rated.

The Lombard effect in children with cochlear implants: suprasegmental aspects

ABSTRACT Children with cochlear implants (CI) communicate in noisy environments, such as in classrooms, where multiple talkers and reverberation are present. Speakers compensate for noise via the ‘Lombard effect’. The present study examined the Lombard effect on the intensity and duration of stressed vowels in the speech of children with Cochlear Implants (CIs) as compared to children with Normal Hearing (NH), focusing on the effects of speech-shaped noise (SSN) and speech-shaped noise with reverberation (SSN+Reverberation). The sample consisted of 7 children with CIs and 7 children with NH, aged 7–12 years. Regarding intensity, a) children with CIs produced stressed vowels with an overall greater intensity across acoustic conditions as compared to NH peers, b) both groups increased their stressed vowel intensity for all vowels from Quiet to both noise conditions, and c) children with NH further increased their intensity when reverberation was added to SSN, esp. for the vowel /u/. Regarding duration, longer stressed vowels were produced by children with CIs as compared to NH in Quiet and SSN conditions but the effect was retained only for the vowels /i/, /o/ and /u/ when reverberation was added to noise. The SSN+Reverberation condition induced systematic lengthening in stressed vowels for children with NH. Furthermore, although greater intensity and duration ratios of stressed/unstressed syllables were observed for children with NH as compared to CIs in Quiet condition, they diminished with noise. The differences observed across groups have implications for speaking in classroom noise.

Individual differences in talker discrimination among adult cochlear implant users

Real-world speech communication involves interacting with talkers with diverse voices and accents. Adult cochlear implant (CI) users often demonstrate poor discrimination of talkers’ voices relative to their normal-hearing peers, which may contribute to difficulties in understanding speech produced by multiple talkers. However, the factors underlying poor talker discrimination ability are not well understood. The current study examined auditory and cognitive-linguistic factors that support talker discrimination, and explored the association between talker discrimination and sentence recognition across CI users. In an AX talker discrimination task, 25 adult CI users indicated whether word pairs were produced by the same talker or by two different talkers. Participants also completed measures of spectrotemporal processing, cognitive-linguistic skills, including fluid intelligence, working memory capacity, inhibitory control, and phonological processing, and sentence recognition. Results showed that talker discrimination scores (accuracy, sensitivity) were moderately to strongly correlated with spectrotemporal processing, fluid intelligence, inhibitory control, and phonological processing. Talker discrimination scores were also related to sentence recognition accuracy scores across CI users. These preliminary findings suggest that both auditory and cognitive-linguistic processes support talker discrimination, and, furthermore, may underlie the relationship between talker discrimination and sentence recognition in adult CI users.

Binaural fusion and vowel perception in cochlear implant users

Excess binaural fusion can be a problem, especially in noisy environments when speech from multiple talkers are fused together. Previously, we showed that hearing aid users experience abnormally broad binaural fusion that leads to fusion and averaging of dichotic vowels of fundamental frequency (F0) differing by up to 1 octave (Reiss and Molis 2021). This study investigated whether cochlear implant (CI) listeners also experience abnormal fusion and averaging of double vowels in the free field. Six adult CI users (4 females) and four NH adults (1 female) were tested. Stimuli were synthetic vowels/æ/, /a/, /u/ and /i/, at F0 = 106.9, 151.2, and 201.8 Hz. Two different vowels were presented simultaneously from two loudspeakers at ±60°, with same or different F0. Subjects were instructed to select which one or two vowels they perceived. When ΔF0 = 0, NH and CI users often fused and identified only one vowel. As ΔF0 increased, NH listeners increased identification of two vowels, while CI users continued to fuse vowels. The findings show that in realistic free field listening conditions, CI users can experience excess fusion of multiple vowels of differing F0s, presenting additional challenges in noisy listening environments. [Work supported by NIH R01DC013307 and ASA SUREIA program.]

Soundscape, attention and cognitive load

The cocktail party problem was first coined by E Colin Cherry in 1953 and it describes the archetypal challenge of listening in a complex soundscape (e.g., multiple talkers conversing vying for your attention in a crowded restaurant). In the past two decades, the field of psychoacoustics has steadily marched towards understanding how we listen in these naturalistic environments. Many studies have focused on studying the psychological and physiological aspects of auditory scene analysis and object-based attention as well as how these processes differ in typical listeners from others with listening difficulties. In recent years, there has also been a burgeoning interest in understanding how cognitive load and listening effort affect our ease of listening in different situational contexts (e.g., talking while driving). In this talk, a brief survey of modern psychoacoustic experimental approaches will be presented in the hope to spur new collaborative research ideas with those who study soundscape.

Development of a feline model for preclinical research of a new translabyrinthine auditory nerve implant.

Cochlear implants are among the most successful neural prosthetic devices to date but exhibit poor frequency selectivity and the inability to consistently activate apical (low frequency) spiral ganglion neurons. These issues can limit hearing performance in many cochlear implant patients, especially for understanding speech in noisy environments and in perceiving or appreciating more complex inputs such as music and multiple talkers. For cochlear implants, electrical current must pass through the bony wall of the cochlea, leading to widespread activation of auditory nerve fibers. Cochlear implants also cannot be implanted in some individuals with an obstruction or severe malformations of the cochlea. Alternatively, intraneural stimulation delivered via an auditory nerve implant could provide direct contact with neural fibers and thus reduce unwanted current spread. More confined current during stimulation can increase selectivity of frequency fiber activation. Furthermore, devices such as the Utah Slanted Electrode Array can provide access to the full cross section of the auditory nerve, including low frequency fibers that are difficult to reach using a cochlear implant. However, further scientific and preclinical research of these Utah Slanted Electrode Array devices is limited by the lack of a chronic large animal model for the auditory nerve implant, especially one that leverages an appropriate surgical approach relevant for human translation. This paper presents a newly developed transbullar translabyrinthine surgical approach for implanting the auditory nerve implant into the cat auditory nerve. In our first of a series of studies, we demonstrate a surgical approach in non-recovery experiments that enables implantation of the auditory nerve implant into the auditory nerve, without damaging the device and enabling effective activation of the auditory nerve fibers, as measured by electrode impedances and electrically evoked auditory brainstem responses. These positive results motivate performing future chronic cat studies to assess the long-term stability and function of these auditory nerve implant devices, as well as development of novel stimulation strategies that can be translated to human patients.

Phonological discrimination and contrast detection in pupillometry.

The perception of phonemes is guided by both low-level acoustic cues and high-level linguistic context. However, differentiating between these two types of processing can be challenging. In this study, we explore the utility of pupillometry as a tool to investigate both low- and high-level processing of phonological stimuli, with a particular focus on its ability to capture novelty detection and cognitive processing during speech perception. Pupillometric traces were recorded from a sample of 22 Danish-speaking adults, with self-reported normal hearing, while performing two phonological-contrast perception tasks: a nonword discrimination task, which included minimal-pair combinations specific to the Danish language, and a nonword detection task involving the detection of phonologically modified words within sentences. The study explored the perception of contrasts in both unprocessed speech and degraded speech input, processed with a vocoder. No difference in peak pupil dilation was observed when the contrast occurred between two isolated nonwords in the nonword discrimination task. For unprocessed speech, higher peak pupil dilations were measured when phonologically modified words were detected within a sentence compared to sentences without the nonwords. For vocoded speech, higher peak pupil dilation was observed for sentence stimuli, but not for the isolated nonwords, although performance decreased similarly for both tasks. Our findings demonstrate the complexity of pupil dynamics in the presence of acoustic and phonological manipulation. Pupil responses seemed to reflect higher-level cognitive and lexical processing related to phonological perception rather than low-level perception of acoustic cues. However, the incorporation of multiple talkers in the stimuli, coupled with the relatively low task complexity, may have affected the pupil dilation.

Classification of head movements during multi-person conversation

Hearing-aid processing strategies could be greatly improved by real-time knowledge of the users’ listening intentions, information that may be encoded by characteristic head movements during conversational turn-takingwith multiple talkers, and measured by accelerometers in the hearing devices. First, however, it must be determined to what degree head movements are stereotypic across listeners, or alternatively, unique to each individual, and whether an automatic classifier can identify specific movements. Three cohorts of three young, normal-hearing individuals participated in a semi-structured conversation for 50 minutes. Participants wore hearing aids with accelerometers and were video-recorded. Video recordings were then used to track head movements using zFace (Jeni et al. 2017) and manually annotate head movement and communication activities. Accelerometer data was segmented into 1-s windows and labeled by the annotated activity identified in that window. Features based on temporal and frequency characteristics were used to train multi-class models with across-subject or within-subject design. Significantly better than chance classification was observed for all activities using both model designs, with higher classification accuracy for the models trained on the individuals’ own-data. In summary, head movement behaviors can be classified with short-scale time resolution, and individual differences are a key factor for accurate classification.

Word Learning in Deaf Adults Who Use Cochlear Implants: The Role of Talker Variability and Attention to the Mouth.

Although cochlear implants (CIs) facilitate spoken language acquisition, many CI listeners experience difficulty learning new words. Studies have shown that highly variable stimulus input and audiovisual cues improve speech perception in CI listeners. However, less is known whether these two factors improve perception in a word learning context. Furthermore, few studies have examined how CI listeners direct their gaze to efficiently capture visual information available on a talker's face. The purpose of this study was two-fold: (1) to examine whether talker variability could improve word learning in CI listeners and (2) to examine how CI listeners direct their gaze while viewing a talker speak. Eighteen adults with CIs and 10 adults with normal hearing (NH) learned eight novel word-object pairs spoken by a single talker or six different talkers (multiple talkers). The word learning task comprised of nonsense words following the phonotactic rules of English. Learning was probed using a novel talker in a two-alternative forced-choice eye gaze task. Learners' eye movements to the mouth and the target object (accuracy) were tracked over time. Both groups performed near ceiling during the test phase, regardless of whether they learned from the same talker or different talkers. However, compared to listeners with NH, CI listeners directed their gaze significantly more to the talker's mouth while learning the words. Unlike NH listeners who can successfully learn words without focusing on the talker's mouth, CI listeners tended to direct their gaze to the talker's mouth, which may facilitate learning. This finding is consistent with the hypothesis that CI listeners use a visual processing strategy that efficiently captures redundant audiovisual speech cues available at the mouth. Due to ceiling effects, however, it is unclear whether talker variability facilitated word learning for adult CI listeners, an issue that should be addressed in future work using more difficult listening conditions.

Cognitive factors in nonnative phonetic learning: Impacts of inhibitory control and working memory on the benefits and costs of talker variability

Talker variability has been reported to facilitate generalization and retention of speech learning, but is also shown to place demands on cognitive resources. Our recent study provided evidence that phonetically-irrelevant acoustic variability in single-talker (ST) speech is sufficient to induce equivalent amounts of learning to the use of multiple-talker (MT) training. This study is a follow-up contrasting MT versus ST training with varying degrees of temporal exaggeration to examine how cognitive measures of individual learners may influence the role of input variability in immediate learning and long-term retention. Native Chinese-speaking adults were trained on the English /i/-/ɪ/ contrast. We assessed the trainees’ working memory and inhibition control before training. The two trained groups showed comparable long-term retention of training effects in terms of word identification performance and more native-like cue weighting in both perception and production regardless of talker variability condition. The results demonstrate the role of phonetically-irrelevant variability in robust speech learning and modulatory functions of nonlinguistic domain-general inhibitory control and working memory, highlighting the necessity to consider the interaction between input characteristics, task difficulty, and individual differences in cognitive abilities in assessing learning outcomes.

Talker adaptation or "talker" adaptation? Musical instrument variability impedes pitch perception.

Listeners show perceptual benefits (faster and/or more accurate responses) when perceiving speech spoken by a single talker versus multiple talkers, known as talker adaptation.While near-exclusively studied in speech and with talkers, some aspects of talker adaptation might reflect domain-general processes. Music, like speech, is a sound class replete with acoustic variation, such as a multitude of pitch and instrument possibilities. Thus, it was hypothesized that perceptual benefits from structure in the acoustic signal (i.e., hearing the same sound source on every trial) are not specific to speech but rather a general auditory response. Forty nonmusician participants completed a simple musical task that mirrored talker adaptation paradigms. Low- or high-pitched notes were presented in single- and mixed-instrument blocks. Reflecting both music research on pitch and timbre interdependence and mirroring traditional "talker" adaptation paradigms, listeners were faster to make their pitch judgments when presented with a single instrument timbre relative to when the timbre was selected from one of four instruments from trial to trial. A second experiment ruled out the possibility that participants were responding faster to the specific instrument chosen as the single-instrument timbre. Consistent with general theoretical approaches to perception, perceptual benefits from signal structure are not limited to speech.

Influences of talker variability, reading ability, and language ability on word learning

Studies show that talker variability is beneficial for learning second language contrasts, though less is known about variability in processing speech in a speaker’s native language. Talker variability benefits have also been found for word learning studies with children, with better performance for children in the high variability condition than those in the low variability condition. Furthermore, it has been shown that individuals with a range of reading and language abilities perform differently on tasks that tap into phonological processes. Therefore, the current study examines how individual differences in reading and language abilities influence word learning and the extent to which talker variability improves performance. One hundred and forty listeners were randomly assigned to three conditions: (1) single talker, (2) multiple talker (randomized), and (3) multiple talker (blocked). Our results showed that individuals with poorer reading or language skills performed worse than individuals with higher reading and language skills. Surprisingly, variability was not beneficial and instead, listeners in the single talker condition performed better than those in the multiple talker conditions. These results suggest that talker variability benefits in word learning may be dependent on other aspects such as the amount of difference between talkers or the task.

Short Implicit Voice Training Affects Listening Effort During a Voice Cue Sensitivity Task With Vocoder-Degraded Speech.

Understanding speech in real life can be challenging and effortful, such as in multiple-talker listening conditions. Fundamental frequency ( fo ) and vocal-tract length ( vtl ) voice cues can help listeners segregate between talkers, enhancing speech perception in adverse listening conditions. Previous research showed lower sensitivity to fo and vtl voice cues when speech signal was degraded, such as in cochlear implant hearing and vocoder-listening compared to normal hearing, likely contributing to difficulties in understanding speech in adverse listening. Nevertheless, when multiple talkers are present, familiarity with a talker's voice, via training or exposure, could provide a speech intelligibility benefit. In this study, the objective was to assess how an implicit short-term voice training could affect perceptual discrimination of voice cues ( fo+vtl ), measured in sensitivity and listening effort, with or without vocoder degradations. Voice training was provided via listening to a recording of a book segment for approximately 30 min, and answering text-related questions, to ensure engagement. Just-noticeable differences (JNDs) for fo+vtl were measured with an odd-one-out task implemented as a 3-alternative forced-choice adaptive paradigm, while simultaneously collecting pupil data. The reference voice either belonged to the trained voice or an untrained voice. Effects of voice training (trained and untrained voice), vocoding (non-vocoded and vocoded), and item variability (fixed or variable consonant-vowel triplets presented across three items) on voice cue sensitivity ( fo+vtl JNDs) and listening effort (pupillometry measurements) were analyzed. Results showed that voice training did not have a significant effect on voice cue discrimination. As expected, fo+vtl JNDs were significantly larger for vocoded conditions than for non-vocoded conditions and with variable item presentations than fixed item presentations. Generalized additive mixed models analysis of pupil dilation over the time course of stimulus presentation showed that pupil dilation was significantly larger during fo+vtl discrimination while listening to untrained voices compared to trained voices, but only for vocoder-degraded speech. Peak pupil dilation was significantly larger for vocoded conditions compared to non-vocoded conditions and variable items increased the pupil baseline relative to fixed items, which could suggest a higher anticipated task difficulty. In this study, even though short voice training did not lead to improved sensitivity to small fo+vtl voice cue differences at the discrimination threshold level, voice training still resulted in reduced listening effort for discrimination among vocoded voice cues.

Multiple sources of acoustic variation affect speech processing efficiency.

Phonetic variability across talkers imposes additional processing costs during speech perception, evident in performance decrements when listening to speech from multiple talkers. However, within-talker phonetic variation is a less well-understood source of variability in speech, and it is unknown how processing costs from within-talker variation compare to those from between-talker variation. Here, listeners performed a speeded word identification task in which three dimensions of variability were factorially manipulated: between-talker variability (single vs multiple talkers), within-talker variability (single vs multiple acoustically distinct recordings per word), and word-choice variability (two- vs six-word choices). All three sources of variability led to reduced speech processing efficiency. Between-talker variability affected both word-identification accuracy and response time, but within-talker variability affected only response time. Furthermore, between-talker variability, but not within-talker variability, had a greater impact when the target phonological contrasts were more similar. Together, these results suggest that natural between- and within-talker variability reflect two distinct magnitudes of common acoustic-phonetic variability: Both affect speech processing efficiency, but they appear to have qualitatively and quantitatively unique effects due to differences in their potential to obscure acoustic-phonemic correspondences across utterances.

Attentive Training: A New Training Framework for Speech Enhancement.

Dealing with speech interference in a speech enhancement system requires either speaker separation or target speaker extraction. Speaker separation has multiple output streams with arbitrary assignments while target speaker extraction requires additional cueing for speaker selection. Both of these are not suitable for a standalone speech enhancement system with one output stream. In this study, we propose a novel training framework, called Attentive Training, to extend speech enhancement to deal with speech interruptions. Attentive training is based on the observation that, in the real world, multiple talkers very unlikely start speaking at the same time, and therefore, a deep neural network can be trained to create a representation of the first speaker and utilize it to attend to or track that speaker in a multitalker noisy mixture. We present experimental results and comparisons to demonstrate the effectiveness of attentive training for speech enhancement.

Talker variability is not always the right noise: 14 month olds struggle to learn dissimilar word-object pairs under talker variability conditions

Seminal work by Stager & Werker (1997) finds that 14-month-olds can rapidly learn two word-object pairings if the words are distinct (e.g. “neem” and “lif”) but not similar (e.g. the minimal pair “bih” and “dih”). More recently, studies have found that adding talker variability during exposure to new word-object pairs lets 14-month-olds succeed on the more challenging minimal pair task, presumably due to talker variability highlighting the “relevant” consistencies between the similar words (Rost & McMurray, 2009; Galle et al., 2015; Hohle et al., 2020). It remains an open question, however, whether talker variability would be similarly useful for learning new word-object pairings when the words themselves are already distinct, or whether instead this extra variability may extinguish learning due to increased task demands. We find evidence for the latter. Namely, in our sample of 54 English-learning 14-month-olds, training infants on two word-object pairings (e.g. “neem” with a dog toy and “lof” with a kitchen tool) only led them to notice when the words and objects were switched if they were trained with single-speaker identical word tokens. When the training featured talker variability (from one or multiple talkers) infants failed to learn the pairings. We suggest that when talker variability is not necessary to highlight the invariant differences between similar words, it may actually increase task difficulty, making it harder for infants to determine what to attend to in the earliest phases of word learning.

Examining the effectiveness of multi-talker training on morphophonological learning

This study explores how auditory stimuli from multiple talkers affects learning of morphophonological alternations in an artificial grammar. Hearing words from multiple voices creates more robust lexical representations (e.g., Davis & Gerken, 2013, 2014; Rost & McMurray, 2009, 2010) and aids generalization to new grammatical structures (Gonzales et al., 2018) and accents (Baese-Berk et al., 2013; Bradlow & Bent, 2008). In ongoing work, participants hear either four different voices or four repetitions from the same voice when learning singular and plural forms in two artificial grammars. As found previously, we expect that hearing from multiple talkers will result in more accurate responses when participants are tested when compared to the participants who heard the same number of repetitions from the same voice. Furthermore, we expect a more pronounced difference in accuracy between the two talker conditions in the Arabic-like grammar than in the English-like grammar.

Hearing Aid Interventions and Its Effect on Hearing Handicap Inventory for Elderly Score Among Elders with Bilateral Hearing Impairment

Background: Several elderly populations face impairment in hearing, difficulty in following discussions in the presence of multiple talkers, understanding the speech of unfamiliar people, understanding fast speech, and understanding speech in noisy environments. Many hearing rehabilitation programs came into the picture for the help of the elderly population. The aim of the meta-analysis was to assess the effect of hearing aid interventions on hearing handicap inventory for elderly (HHIE) scores among elders with bilateral hearing impairment in randomized controlled trials (RCTs). Materials and Methods: Two independent reviewers searched for full-text RCTs from worldwide online databases named PubMed and Google Scholar. Studies from January 2000 to January 2022 in the English language conducted among humans were considered. The search strategy contains PICO format in which P-elders with bilateral hearing impairment, I-hearing aid intervention, C-placebo, O-HHIE scores as outcome measures. Results: A total of 28 trials were identified from online databases using PICO format further on evaluation 8 trials with a total of 445 participants were included for the final meta-analysis of the study. The pooled results after sensitivity analysis revealed that 7 trials had a significant reduction in mean difference (MD) of HHIE score among elders with bilateral hearing impairment (MD-2.21 (95% confidence interval [CI] = −4.24 to −0.19), P = 0.03 with heterogeneity (I 2 = 48%, P = 0.07) in the fixed-effect model. The Stratified analysis was also done between subgroups as HHIE-social/situational and HHIE-emotional scores shows no significant difference among subgroups for the reduction in a MD of scores −5.34 (95% CI −7.05, −3.64), P = 0.03 without any heterogeneity (I 2 = 0%, P = 0.97) with use of fixed-effect model. Conclusion: The findings of the meta-analysis revealed that any type of hearing aids interventional program is found to be effective in reducing HHIE scores. It indicates that these interventions result in a reduction in social and emotional consequences due to hearing loss among elders with bilateral hearing impairment. Adherence to the hearing rehabilitation programs is highly recommended by the researchers to improve social and emotional aspects of life among elders.

Adapting to talkers in speech perception: Balancing generalization and individuation

Listeners can account for systematic variability between talkers, which is learned over exposure to multiple talkers. Previous research suggests that listeners can both generalize prior knowledge of phoneme categories from a familiar to a novel talker (Eisner & McQueen, 2005; Kraljic & Samuel, 2006, 2007) and individuate talkers, preventing generalization (Luthra, Mechtenberg, & Myers, 2021; Tamminga, Wilder, Lai, & Wade, 2020). It is unclear how listeners balance these competing demands. Participants (n = 160) learned two novel talkers (one male and one female voice) with a unique voice onset time (VOT) boundary across two days. On each day, participants were passively exposed to a bimodal distribution of VOTs from one talker, then tested on a second talker (uniform distribution). Day 1 assessed how listeners generalize to a novel talker while Day 2 assessed how the talker that was learned on Day 1 is individuated from the new talker. We found evidence for generalization after Day 1 but little evidence of learning after learning both talkers on Day 2. Two follow-up experiments using interleaved designs and supervised learning also showed little evidence for individuation. This suggests that listeners are likely accounting for variability by shifting their VOT boundary to match current input.