Wearable Vibrotactile Speech Aid

Categories: “Physics, Optics and Electronic Devices” , “Computer Science“, “Medical & Research Devices

Reference #: 2017-043

OTC Contact: Zeinab Abouissa,  Phone: 202-687-2702, Email: zaa9@georgetown.edu


Humans process speech through auditory signals. However, either due to noisy environments or pathology (hearing loss and aging), these auditory signals can be degraded, leading to impaired ability to understand the spoken word. Humans can also communicate by converting auditory signals into haptic signals and can learn to recognize speech through vibrotactile sensations. Still, impractically long training times and mixed outcomes have been a stumbling block for device development in this field.

Researchers at Georgetown University’s Department of Neuroscience developed a novel technology and approach to tackle this problem. This technology is a vibrotactile speech aid that aids hearing-impaired individuals as well as the hearing of normal individuals in noisy conditions such cockpits or construction sites. It does so by converting the auditory signal into vibratory patterns that are designed to optimally interface with the brain’s auditory speech representations. The device reads an incoming auditory speech signal, extracts the salient speech information (either from one or multiple channels) and presents them in a way that is designed to “piggy-back” onto the brain’s auditory speech system. This makes use of the researchers’ experimental results that have shown how the vibrotactile system can efficiently connect to the auditory speech system, provided the vibrotactile signal is formatted in a way that optimally interfaces with the brain’s speech system (Damera et al., 2021), see Figure 1.

The approach further includes a novel auditory-to-vibrotactile conversion algorithm that leverages their years of research into optimally coupling neural processing pathways as well as neuroscience-inspired training protocols. By leveraging novel insights into the mechanisms of speech learning and the coupling of vibrotactile and auditory speech processing pathways in the brain, training time can be dramatically decreased and the efficacy of haptic speech devices can be significantly achieved.

Most vibrotactile speech devices to date have focused on sensory substitution – communicating speech via vibration in the absence of auditory input. In addition, non-intuitive auditory/haptic correspondences and long training times have been major obstacles to adoption. The advantage of the present invention lies in the deep understanding of the neuroscience underlying haptic and speech processing – fields in which the researchers’ group is a world leader. The researchers show that their understanding of how the brain processes sensory stimuli and achieves cross-modal coupling is of critical importance for success in applied neuroscience.

In conclusion, this technology focuses on the coupling of auditory and haptic speech processing and new neuroscience-based approaches to drastically decrease training times and increase usability of the technology. The outcome is a vibrotactile speech aid that can be learned in a fraction of the time of conventional approaches by effectively linking the brain’s haptic system with the speech system.

Figure 1: Illustration of some of the research underlying our
IP and approach. Top left: We use fMRI to identify speech selective
areas in the brain (circle). While these areas are
not selective for vibrotactile (VT) speech before training
(top right, showing two groups of participants, one to-betrained
with the algorithm optimized for VT-to-auditory
speech coupling [vocoded], and one to-be-trained with a
conventional algorithm based on the literature (Reed et al.,
2018)), our optimized speech-to-vibrotactile conversion
algorithm makes the auditory speech system respond to
vibrotactile speech following training (bottom left, circle –
compare to top left). In contrast, no such engagement is
found for the token-based algorithm (bottom right, note no
responses in the auditory speech area).


Max Riesenhuber, Ph.D.
Patrick Malone M.D., Ph.D.


U.S. Patent Application No. 16/643,824
European Patent Application No. 18850514.3