Every year, many thousands of people worldwide lose the ability to speak due to receiving a laryngectomy, typically for treatment of cancer. At some point in their recovery, most will use an electrolarynx to recover their ability to speak. Typical electrolarynxes utilize a piston to strike a disc pressed to the patient’s neck which delivers a pressure wave into the soft tissue. This pressure wave mechanically couples with the vocal tract and generates the fundamental frequency necessary for creating vowels without which speech is not possible.
Commonly available electrolarynxes suffer from poor frequency control due to the nonlinear character of their impulse driver. They also create a great deal of “self-noise” which is distracting to listeners and makes using voice communication systems difficult.
We created a novel electrolarynx implementation which utilizes two interfering ultrasonic waves to generate a fundamental frequency in the vocal tract required for speech restoration. The device is light weight, compact, inexpensive, and offers excellent control of all aspects of the output waveform. In addition, as the primary waveforms are above human hearing, there is little “self-noise” that can be heard by listeners and most communications devices filter such noise as part of their standard digitization process.
This device offers the potential to greatly improve the lives of those who have lost their voices and must rely on technology to allow them to communicate in the most efficient manner.
P M Mills, An Audible Ultrasound Electrolarynx, Ph.D. dissertation, School of Engineering & Applied Science, George Washington University, Washington, DC, 2015.
P Mills and J Zara. “3D Simulation of an Audible Ultrasonic Electrolarynx Using Difference Waves.” PLoS ONE. 9(11): e113339. doi:10.1371/journal.pone.0113339.
P Mills, Defense Presentation, Wednesday, March 26, 2014 at 10:00 a.m.
Please note: Currently none of these videos have any sound; they are purely visual.
AustinMan model – 3D render of the AustinMan Electromagnetic Voxels Model v1.1 Partial Body male model used in our simulations.
2D FDTD simulation – this simulation shows both pressure & velocity waves in both the transverse and sagittal planes.
3D FDTD simulation – shows pressure waves generated by dual low frequency ultrasonic emitters. This 3D simulation was implemented in CUDA, C++ & OpenGL handling 320 1mm slices of 512 mm x 288 mm.
I created a basic phantom for testing the device in the real-world. For comparison purposes, a digital model of the phantom was also created. These videos show the phantom.
video of the phantom & digital phantom model
phantom simulation – a 3D FDTD simulation using the digital phantom model.
During my Qualifying Examination, one of my professors asked why I had used full tissue parameters rather than just assuming soft tissue everywhere. While a soft tissue assumption would have allowed me to speed up the simulation by 2x, we do not get the level of detail required for our results.
Below are three videos that compare a full tissue property simulation to a soft tissue only simulation. It is difficult to see the differences, but watch boney areas such as the spinal region - differences are particularly prominent in the velocity wave view.
Tissue Comparison - Sagittal & Transverse views
Tissue Comparison - Single Source, Transverse view
Contact me at: pmills@ieee.org - © 1995 - 2024 Patrick M. Mills. All Rights Reserved.