eLife assessment: Reconstructing Voice Identity from Noninvasive Auditory Cortex Recordings

Published: July 15, 2024

The cerebral processing of voice information is known to engage, in human as well non-human primates, "temporal areas" (TVAs) that respond preferentially conspecific vocalizations. However, how represented by neuronal populations these areas, particularly speaker identity information, remains poorly understood. Here, we used a deep neural network (DNN) generate high-level, small-dimension representational space for identity—the 'voice latent space' (VLS)—and examined its linear relation with activity via encoding, similarity, and decoding analyses. We find the VLS maps onto fMRI measures response tens thousands stimuli from hundreds different identities better accounts geometry TVAs than A1. Moreover, allowed TVA-based reconstructions preserved essential aspects assessed both machine classifiers listeners. These results indicate DNN-derived provides high-level representations TVAs.

Language: Английский

Reconstructing Voice Identity from Noninvasive Auditory Cortex Recordings

Published: July 15, 2024

The cerebral processing of voice information is known to engage, in human as well non-human primates, “temporal areas” (TVAs) that respond preferentially conspecific vocalizations. However, how represented by neuronal populations these areas, particularly speaker identity information, remains poorly understood. Here, we used a deep neural network (DNN) generate high-level, small-dimension representational space for identity—the ‘voice latent space’ (VLS)—and examined its linear relation with activity via encoding, similarity, and decoding analyses. We find the VLS maps onto fMRI measures response tens thousands stimuli from hundreds different identities better accounts geometry TVAs than A1. Moreover, allowed TVA-based reconstructions preserved essential aspects assessed both machine classifiers listeners. These results indicate DNN-derived provides high-level representations TVAs.

Language: Английский

Functional connectivity across the human subcortical auditory system using a Gaussian copula graphical model approach with partial correlations

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2022, Volume and Issue: unknown

Published: Sept. 17, 2022

Abstract/Summary The auditory system comprises multiple subcortical brain structures that process and refine incoming acoustic signals along the primary pathway. Due to technical limitations of imaging small deep inside brain, most our knowledge is based on research in animal models using invasive methodologies. Advances ultra-high field functional magnetic resonance (fMRI) acquisition have enabled novel non-invasive investigations human subcortex, including fundamental features representation such as tonotopy periodotopy. However, connectivity across networks still underexplored humans, with ongoing development related methods. Traditionally, estimated from fMRI data full correlation matrices. partial correlations reveal relationship between two regions after removing effects all other regions, reflecting more direct connectivity. Partial analysis particularly promising ascending system, where sensory information passed an obligatory manner, nucleus up pathway, providing redundant but also increasingly abstract representations stimuli. While existing methods for learning conditional dependency assume independently identically Gaussian distributed data, exhibit significant deviations Gaussianity well high temporal autocorrelation. In this paper, we developed autoregressive matrix-Gaussian copula graphical model (ARMGCGM) approach estimate thereby infer patterns within while appropriately accounting autocorrelations successive scans. Our results show strong positive pathway each side (left right), midbrain thalamus, associative cortex. These are highly stable when splitting halves according schemes computing separately half cross-validation folds. contrast, correlation-based identified a rich network interconnectivity was not specific adjacent nodes Overall, demonstrate unique recoverable approaches reliable acquisitions.

Language: Английский