An open-source toolbox for measuring vocal tract shape from real-time magnetic resonance images.

Real-time magnetic resonance imaging (rtMRI) is a technique that provides high-contrast videographic data of human anatomy in motion. Applied to the vocal tract, it is a powerful method for capturing the dynamics of speech and other vocal behaviours by imaging structures internal to the mouth and throat. These images provide a means of studying the physiological basis for speech, singing, expressions of emotion, and swallowing that are otherwise not accessible for external observation. However, taking quantitative measurements from these images is notoriously difficult. We introduce a signal processing pipeline that produces outlines of the vocal tract from the lips to the larynx as a quantification of the dynamic morphology of the vocal tract. Our approach performs simple tissue classification, but constrained to a researcher-specified region of interest. This combination facilitates feature extraction while retaining the domain-specific expertise of a human analyst. We demonstrate that this pipeline generalises well across datasets covering behaviours such as speech, vocal size exaggeration, laughter, and whistling, as well as producing reliable outcomes across analysts, particularly among users with domain-specific expertise. With this article, we make this pipeline available for immediate use by the research community, and further suggest that it may contribute to the continued development of fully automated methods based on deep learning algorithms.

Talker and accent familiarity yield advantages for voice identity perception: A voice sorting study.

In the current study, we examine and compare the effects of talker and accent familiarity in the context of a voice identity sorting task, using naturally varying voice recording samples from the TV show Derry Girls. Voice samples were thus all spoken with a regional accent of UK/Irish English (from [London]derry). We tested four listener groups: Listeners were either familiar or unfamiliar with the TV show (and therefore the talker identities) and were either highly familiar or relatively less familiar with Northern Irish accents. Both talker and accent familiarity significantly improved accuracy of voice identity sorting performance. However, the talker familiarity benefits were overall larger, and more consistent. We discuss the results in light of a possible hierarchy of familiarity effects and argue that our findings may provide additional evidence for interactions of speech and identity processing pathways in voice identity perception. We also identify some key limitations in the current work and provide suggestions for future studies to address these.

FFA and OFA Encode Distinct Types of Face Identity Information.

Faces of different people elicit distinct fMRI patterns in several face-selective regions of the human brain. Here we used representational similarity analysis to investigate what type of identity-distinguishing information is encoded in three face-selective regions: fusiform face area (FFA), occipital face area (OFA), and posterior superior temporal sulcus (pSTS). In a sample of 30 human participants (22 females, 8 males), we used fMRI to measure brain activity patterns elicited by naturalistic videos of famous face identities, and compared their representational distances in each region with models of the differences between identities. We built diverse candidate models, ranging from low-level image-computable properties (pixel-wise, GIST, and Gabor-Jet dissimilarities), through higher-level image-computable descriptions (OpenFace deep neural network, trained to cluster faces by identity), to complex human-rated properties (perceived similarity, social traits, and gender). We found marked differences in the information represented by the FFA and OFA. Dissimilarities between face identities in FFA were accounted for by differences in perceived similarity, Social Traits, Gender, and by the OpenFace network. In contrast, representational distances in OFA were mainly driven by differences in low-level image-based properties (pixel-wise and Gabor-Jet dissimilarities). Our results suggest that, although FFA and OFA can both discriminate between identities, the FFA representation is further removed from the image, encoding higher-level perceptual and social face information.SIGNIFICANCE STATEMENT Recent studies using fMRI have shown that several face-responsive brain regions can distinguish between different face identities. It is however unclear whether these different face-responsive regions distinguish between identities in similar or different ways. We used representational similarity analysis to investigate the computations within three brain regions in response to naturalistically varying videos of face identities. Our results revealed that two regions, the fusiform face area and the occipital face area, encode distinct identity information about faces. Although identity can be decoded from both regions, identity representations in fusiform face area primarily contained information about social traits, gender, and high-level visual features, whereas occipital face area primarily represented lower-level image features.

Unimodal and cross-modal identity judgements using an audio-visual sorting task: Evidence for independent processing of faces and voices.

Unimodal and cross-modal information provided by faces and voices contribute to identity percepts. To examine how these sources of information interact, we devised a novel audio-visual sorting task in which participants were required to group video-only and audio-only clips into two identities. In a series of three experiments, we show that unimodal face and voice sorting were more accurate than cross-modal sorting: While face sorting was consistently most accurate followed by voice sorting, cross-modal sorting was at chancel level or below. In Experiment 1, we compared performance in our novel audio-visual sorting task to a traditional identity matching task, showing that unimodal and cross-modal identity perception were overall moderately more accurate than the traditional identity matching task. In Experiment 2, separating unimodal from cross-modal sorting led to small improvements in accuracy for unimodal sorting, but no change in cross-modal sorting performance. In Experiment 3, we explored the effect of minimal audio-visual training: Participants were shown a clip of the two identities in conversation prior to completing the sorting task. This led to small, nonsignificant improvements in accuracy for unimodal and cross-modal sorting. Our results indicate that unfamiliar face and voice perception operate relatively independently with no evidence of mutual benefit, suggesting that extracting reliable cross-modal identity information is challenging.

Human larynx motor cortices coordinate respiration for vocal-motor control.

Vocal flexibility is a hallmark of the human species, most particularly the capacity to speak and sing. This ability is supported in part by the evolution of a direct neural pathway linking the motor cortex to the brainstem nucleus that controls the larynx the primary sound source for communication. Early brain imaging studies demonstrated that larynx motor cortex at the dorsal end of the orofacial division of motor cortex (dLMC) integrated laryngeal and respiratory control, thereby coordinating two major muscular systems that are necessary for vocalization. Neurosurgical studies have since demonstrated the existence of a second larynx motor area at the ventral extent of the orofacial motor division (vLMC) of motor cortex. The vLMC has been presumed to be less relevant to speech motor control, but its functional role remains unknown. We employed a novel ultra-high field (7T) magnetic resonance imaging paradigm that combined singing and whistling simple melodies to localise the larynx motor cortices and test their involvement in respiratory motor control. Surprisingly, whistling activated both 'larynx areas' more strongly than singing despite the reduced involvement of the larynx during whistling. We provide further evidence for the existence of two larynx motor areas in the human brain, and the first evidence that laryngeal-respiratory integration is a shared property of both larynx motor areas. We outline explicit predictions about the descending motor pathways that give these cortical areas access to both the laryngeal and respiratory systems and discuss the implications for the evolution of speech.

Vocomotor and Social Brain Networks Work Together to Express Social Traits in Voices.

Voice modulation is important when navigating social interactions-tone of voice in a business negotiation is very different from that used to comfort an upset child. While voluntary vocal behavior relies on a cortical vocomotor network, social voice modulation may require additional social cognitive processing. Using functional magnetic resonance imaging, we investigated the neural basis for social vocal control and whether it involves an interplay of vocal control and social processing networks. Twenty-four healthy adult participants modulated their voice to express social traits along the dimensions of the social trait space (affiliation and competence) or to express body size (control for vocal flexibility). Naïve listener ratings showed that vocal modulations were effective in evoking social trait ratings along the two primary dimensions of the social trait space. Whereas basic vocal modulation engaged the vocomotor network, social voice modulation specifically engaged social processing regions including the medial prefrontal cortex, superior temporal sulcus, and precuneus. Moreover, these regions showed task-relevant modulations in functional connectivity to the left inferior frontal gyrus, a core vocomotor control network area. These findings highlight the impact of the integration of vocal motor control and social information processing for socially meaningful voice modulation.

Faces and voices in the brain: A modality-general person-identity representation in superior temporal sulcus.

Face-selective and voice-selective brain regions have been shown to represent face-identity and voice-identity, respectively. Here we investigated whether there are modality-general person-identity representations in the brain that can be driven by either a face or a voice, and that invariantly represent naturalistically varying face videos and voice recordings of the same identity. Models of face and voice integration suggest that such representations could exist in multimodal brain regions, and in unimodal regions via direct coupling between face- and voice-selective regions. Therefore, in this study we used fMRI to measure brain activity patterns elicited by the faces and voices of familiar people in face-selective, voice-selective, and person-selective multimodal brain regions. We used representational similarity analysis to (1) compare representational geometries (i.e. representational dissimilarity matrices) of face- and voice-elicited identities, and to (2) investigate the degree to which pattern discriminants for pairs of identities generalise from one modality to the other. We did not find any evidence of similar representational geometries across modalities in any of our regions of interest. However, our results showed that pattern discriminants that were trained to discriminate pairs of identities from their faces could also discriminate the respective voices (and vice-versa) in the right posterior superior temporal sulcus (rpSTS). Our findings suggest that the rpSTS is a person-selective multimodal region that shows a modality-general person-identity representation and integrates face and voice identity information.

Vocal Tract Images Reveal Neural Representations of Sensorimotor Transformation During Speech Imitation.

Imitating speech necessitates the transformation from sensory targets to vocal tract motor output, yet little is known about the representational basis of this process in the human brain. Here, we address this question by using real-time MR imaging (rtMRI) of the vocal tract and functional MRI (fMRI) of the brain in a speech imitation paradigm. Participants trained on imitating a native vowel and a similar nonnative vowel that required lip rounding. Later, participants imitated these vowels and an untrained vowel pair during separate fMRI and rtMRI runs. Univariate fMRI analyses revealed that regions including left inferior frontal gyrus were more active during sensorimotor transformation (ST) and production of nonnative vowels, compared with native vowels; further, ST for nonnative vowels activated somatomotor cortex bilaterally, compared with ST of native vowels. Using test representational similarity analysis (RSA) models constructed from participants' vocal tract images and from stimulus formant distances, we found that RSA searchlight analyses of fMRI data showed either type of model could be represented in somatomotor, temporal, cerebellar, and hippocampal neural activation patterns during ST. We thus provide the first evidence of widespread and robust cortical and subcortical neural representation of vocal tract and/or formant parameters, during prearticulatory ST.

Cohesion and Joint Speech: Right Hemisphere Contributions to Synchronized Vocal Production.

UNLABELLED: Synchronized behavior (chanting, singing, praying, dancing) is found in all human cultures and is central to religious, military, and political activities, which require people to act collaboratively and cohesively; however, we know little about the neural underpinnings of many kinds of synchronous behavior (e.g., vocal behavior) or its role in establishing and maintaining group cohesion. In the present study, we measured neural activity using fMRI while participants spoke simultaneously with another person. We manipulated whether the couple spoke the same sentence (allowing synchrony) or different sentences (preventing synchrony), and also whether the voice the participant heard was "live" (allowing rich reciprocal interaction) or prerecorded (with no such mutual influence). Synchronous speech was associated with increased activity in posterior and anterior auditory fields. When, and only when, participants spoke with a partner who was both synchronous and "live," we observed a lack of the suppression of auditory cortex, which is commonly seen as a neural correlate of speech production. Instead, auditory cortex responded as though it were processing another talker's speech. Our results suggest that detecting synchrony leads to a change in the perceptual consequences of one's own actions: they are processed as though they were other-, rather than self-produced. This may contribute to our understanding of synchronized behavior as a group-bonding tool. SIGNIFICANCE STATEMENT: Synchronized human behavior, such as chanting, dancing, and singing, are cultural universals with functional significance: these activities increase group cohesion and cause participants to like each other and behave more prosocially toward each other. Here we use fMRI brain imaging to investigate the neural basis of one common form of cohesive synchronized behavior: joint speaking (e.g., the synchronous speech seen in chants, prayers, pledges). Results showed that joint speech recruits additional right hemisphere regions outside the classic speech production network. Additionally, we found that a neural marker of self-produced speech, suppression of sensory cortices, did not occur during joint synchronized speech, suggesting that joint synchronized behavior may alter self-other distinctions in sensory processing.

Functional brain outcomes of L2 speech learning emerge during sensorimotor transformation.

Sensorimotor transformation (ST) may be a critical process in mapping perceived speech input onto non-native (L2) phonemes, in support of subsequent speech production. Yet, little is known concerning the role of ST with respect to L2 speech, particularly where learned L2 phones (e.g., vowels) must be produced in more complex lexical contexts (e.g., multi-syllabic words). Here, we charted the behavioral and neural outcomes of producing trained L2 vowels at word level, using a speech imitation paradigm and functional MRI. We asked whether participants would be able to faithfully imitate trained L2 vowels when they occurred in non-words of varying complexity (one or three syllables). Moreover, we related individual differences in imitation success during training to BOLD activation during ST (i.e., pre-imitation listening), and during later imitation. We predicted that superior temporal and peri-Sylvian speech regions would show increased activation as a function of item complexity and non-nativeness of vowels, during ST. We further anticipated that pre-scan acoustic learning performance would predict BOLD activation for non-native (vs. native) speech during ST and imitation. We found individual differences in imitation success for training on the non-native vowel tokens in isolation; these were preserved in a subsequent task, during imitation of mono- and trisyllabic words containing those vowels. fMRI data revealed a widespread network involved in ST, modulated by both vowel nativeness and utterance complexity: superior temporal activation increased monotonically with complexity, showing greater activation for non-native than native vowels when presented in isolation and in trisyllables, but not in monosyllables. Individual differences analyses showed that learning versus lack of improvement on the non-native vowel during pre-scan training predicted increased ST activation for non-native compared with native items, at insular cortex, pre-SMA/SMA, and cerebellum. Our results hold implications for the importance of ST as a process underlying successful imitation of non-native speech.

Getting the Cocktail Party Started: Masking Effects in Speech Perception.

Spoken conversations typically take place in noisy environments, and different kinds of masking sounds place differing demands on cognitive resources. Previous studies, examining the modulation of neural activity associated with the properties of competing sounds, have shown that additional speech streams engage the superior temporal gyrus. However, the absence of a condition in which target speech was heard without additional masking made it difficult to identify brain networks specific to masking and to ascertain the extent to which competing speech was processed equivalently to target speech. In this study, we scanned young healthy adults with continuous fMRI, while they listened to stories masked by sounds that differed in their similarity to speech. We show that auditory attention and control networks are activated during attentive listening to masked speech in the absence of an overt behavioral task. We demonstrate that competing speech is processed predominantly in the left hemisphere within the same pathway as target speech but is not treated equivalently within that stream and that individuals who perform better in speech in noise tasks activate the left mid-posterior superior temporal gyrus more. Finally, we identify neural responses associated with the onset of sounds in the auditory environment; activity was found within right lateralized frontal regions consistent with a phasic alerting response. Taken together, these results provide a comprehensive account of the neural processes involved in listening in noise.

Feel the Noise: Relating Individual Differences in Auditory Imagery to the Structure and Function of Sensorimotor Systems.

Humans can generate mental auditory images of voices or songs, sometimes perceiving them almost as vividly as perceptual experiences. The functional networks supporting auditory imagery have been described, but less is known about the systems associated with interindividual differences in auditory imagery. Combining voxel-based morphometry and fMRI, we examined the structural basis of interindividual differences in how auditory images are subjectively perceived, and explored associations between auditory imagery, sensory-based processing, and visual imagery. Vividness of auditory imagery correlated with gray matter volume in the supplementary motor area (SMA), parietal cortex, medial superior frontal gyrus, and middle frontal gyrus. An analysis of functional responses to different types of human vocalizations revealed that the SMA and parietal sites that predict imagery are also modulated by sound type. Using representational similarity analysis, we found that higher representational specificity of heard sounds in SMA predicts vividness of imagery, indicating a mechanistic link between sensory- and imagery-based processing in sensorimotor cortex. Vividness of imagery in the visual domain also correlated with SMA structure, and with auditory imagery scores. Altogether, these findings provide evidence for a signature of imagery in brain structure, and highlight a common role of perceptual-motor interactions for processing heard and internally generated auditory information.

I thought that I heard you laughing: Contextual facial expressions modulate the perception of authentic laughter and crying.

It is well established that categorising the emotional content of facial expressions may differ depending on contextual information. Whether this malleability is observed in the auditory domain and in genuine emotion expressions is poorly explored. We examined the perception of authentic laughter and crying in the context of happy, neutral and sad facial expressions. Participants rated the vocalisations on separate unipolar scales of happiness and sadness and on arousal. Although they were instructed to focus exclusively on the vocalisations, consistent context effects were found: For both laughter and crying, emotion judgements were shifted towards the information expressed by the face. These modulations were independent of response latencies and were larger for more emotionally ambiguous vocalisations. No effects of context were found for arousal ratings. These findings suggest that the automatic encoding of contextual information during emotion perception generalises across modalities, to purely non-verbal vocalisations, and is not confined to acted expressions.

Exploring the roles of spectral detail and intonation contour in speech intelligibility: an FMRI study.

The melodic contour of speech forms an important perceptual aspect of tonal and nontonal languages and an important limiting factor on the intelligibility of speech heard through a cochlear implant. Previous work exploring the neural correlates of speech comprehension identified a left-dominant pathway in the temporal lobes supporting the extraction of an intelligible linguistic message, whereas the right anterior temporal lobe showed an overall preference for signals clearly conveying dynamic pitch information [Johnsrude, I. S., Penhune, V. B., & Zatorre, R. J. Functional specificity in the right human auditory cortex for perceiving pitch direction. Brain, 123, 155-163, 2000; Scott, S. K., Blank, C. C., Rosen, S., & Wise, R. J. Identification of a pathway for intelligible speech in the left temporal lobe. Brain, 123, 2400-2406, 2000]. The current study combined modulations of overall intelligibility (through vocoding and spectral inversion) with a manipulation of pitch contour (normal vs. falling) to investigate the processing of spoken sentences in functional MRI. Our overall findings replicate and extend those of Scott et al. [Scott, S. K., Blank, C. C., Rosen, S., & Wise, R. J. Identification of a pathway for intelligible speech in the left temporal lobe. Brain, 123, 2400-2406, 2000], where greater sentence intelligibility was predominately associated with increased activity in the left STS, and the greatest response to normal sentence melody was found in right superior temporal gyrus. These data suggest a spatial distinction between brain areas associated with intelligibility and those involved in the processing of dynamic pitch information in speech. By including a set of complexity-matched unintelligible conditions created by spectral inversion, this is additionally the first study reporting a fully factorial exploration of spectrotemporal complexity and spectral inversion as they relate to the neural processing of speech intelligibility. Perhaps surprisingly, there was little evidence for an interaction between the two factors-we discuss the implications for the processing of sound and speech in the dorsolateral temporal lobes.

A little more conversation, a little less action--candidate roles for the motor cortex in speech perception.

The motor theory of speech perception assumes that activation of the motor system is essential in the perception of speech. However, deficits in speech perception and comprehension do not arise from damage that is restricted to the motor cortex, few functional imaging studies reveal activity in the motor cortex during speech perception, and the motor cortex is strongly activated by many different sound categories. Here, we evaluate alternative roles for the motor cortex in spoken communication and suggest a specific role in sensorimotor processing in conversation. We argue that motor cortex activation is essential in joint speech, particularly for the timing of turn taking.

Voluntary and involuntary processes affect the production of verbal and non-verbal signals by the human voice.

We argue that a comprehensive model of human vocal behaviour must address both voluntary and involuntary aspects of articulate speech and non-verbal vocalizations. Within this, plasticity of vocal output should be acknowledged and explained as part of the mature speech production system.

Self-ownership, not self-production, modulates bias and agency over a synthesised voice.

Voices are fundamentally social stimuli, and their importance to the self may be underpinned by how far they can be used to express the self and achieve communicative goals. This paper examines how self-bias and agency over a synthesised voice is altered when that voice is used to represent the self in social interaction. To enable participants to use a new voice, a novel two-player game was created, in which participants communicated online using a text-to-speech (TTS) synthesised voice. We then measured self-bias and sense of agency attributed to this synthesised voice, comparing participants who had used their new voice to interact with another person (n = 44) to a control group of participants (n = 44) who had been only briefly exposed to the voices. We predicted that the new, synthesised self-voice would be more perceptually prioritised after it had been self-produced. Further, that participants' sense of agency over the voice would be increased, if they had experienced self-producing the voice, relative to those who only owned it. Contrary to the hypothesis, the results indicated that both experimental participants and control participants similarly prioritised the new synthesised voice and experienced a similar degree of agency over it, relative to voices owned by others. Critically then, being able to produce the new voice in a social interaction did not modulate bias towards it nor participant's sense of agency over it. These results suggest that merely having ownership over a new voice may be sufficient to generate a perceptual bias and a sense of agency over it.

Sensorimotor learning during synchronous speech is modulated by the acoustics of the other voice.

This study tested the hypothesis that speaking with other voices can influence sensorimotor predictions of one's own voice. Real-time manipulations of auditory feedback were used to drive sensorimotor adaptation in speech, while participants spoke sentences in synchrony with another voice, a task known to induce implicit imitation (phonetic convergence). The acoustic-phonetic properties of the other voice were manipulated between groups, such that convergence with it would either oppose (incongruent group, n = 15) or align with (congruent group, n = 16) speech motor adaptation. As predicted, significantly greater adaptation was seen in the congruent compared to the incongruent group. This suggests the use of shared sensory targets in speech for predicting the sensory outcomes of both the actions of others (speech perception) and the actions of the self (speech production). This finding has important implications for wider theories of shared predictive mechanisms across perception and action, such as active inference.

Human talkers change their voices to elicit specific trait percepts.

The voice is a variable and dynamic social tool with functional relevance for self-presentation, for example, during a job interview or courtship. Talkers adjust their voices flexibly to their situational or social environment. Here, we investigated how effectively intentional voice modulations can evoke trait impressions in listeners (Experiment 1), whether these trait impressions are recognizable (Experiment 2), and whether they meaningfully influence social interactions (Experiment 3). We recorded 40 healthy adult speakers' whilst speaking neutrally and whilst producing vocal expressions of six social traits (e.g., likeability, confidence). Multivariate ratings of 40 listeners showed that vocal modulations amplified specific trait percepts (Experiments 1 and 2), which could be explained by two principal components relating to perceived affiliation and competence. Moreover, vocal modulations increased the likelihood of listeners choosing the voice to be suitable for corresponding social goals (i.e., a confident rather than likeable voice to negotiate a promotion, Experiment 3). These results indicate that talkers modulate their voice along a common trait space for social navigation. Moreover, beyond reactive voice changes, vocal behaviour can be strategically used by talkers to communicate subtle information about themselves to listeners. These findings advance our understanding of non-verbal vocal behaviour for social communication.