Research Feature: VocalTube

The VocalTube project, led by graduate student Debasish Ray Mohapatra, talks about how speech is an essential mechanism for communication and expressing emotions — the words that we pronounce and the nature of expression while speaking define our individuality. Hence, two people never sound similar, even though they speak the same word in the same language. It has been a long fascination in the science community to understand — “how does human produce sound?” and “how could we make machines speak like a human?”. It is clear that we have come a long way from modelling formant speech synthesizers to state-of-the-art powerful machine learning models. However, we are still far away from designing a physics-based articulatory speech synthesizer that could generate speech sound in real-time. Mohapatra’s current published paper, which was shown at the 2019 Interspeech Conference, held in Graz, Austria, addresses this research problem. Currently, he is trying to build

a speaker-specific vocal tract model (2.5D FDTD vocal tract) using the finite difference method that could produce static vowel sounds in quasi-real-time.

Speech production is a complex activity. But in terms of functionality, it is the same as a wind instrument — we blow air through the reed (mouthpiece), which is the VocalTube source of acoustic energy, and those acoustic waves pass through a resonator to produce sound. As we change the geometry of the resonator (duct), the musical sound will vary. In speech anatomy, the non-periodic vibration of vocal folds works as the source and the upper vocal tract as the resonator or articulator. As the articulation (geometry of vocal tract) differs for each individual, we sound very differently while speaking.

But the vocal tract has a very intricate and complex geometry. To capture its irregularity, we need to build a 3D model. That could provide us with better acoustic characteristics, but it’s computationally expensive. Similarly, the existing 1D models provide faster simulation but an oversimplified representation of the realistic vocal tract. We have come across a novel approach for modelling a 2D vocal tract having 3D characteristics using the finite difference time domain (FDTD) numerical method. This new strategy will give us a way to design a real-time speech synthesizer without compromising its acoustic features. And this computational model uses the vocal tract area functions, collected through MRI, while the speaker making vowels and consonants sounds.

Our next goal is to understand the nonlinear coupling between the vocal fold and vocal tract for designing a complete articulatory speech synthesizer, as this research could also be implemented in singing synthesizers and aero-acoustic modelling of wind instruments. Check out the paper and source code below for more information.

Paper Link: https://www.isca-speech.org/archive/Interspeech_2019/pdfs/1764.pdf
Source Code: https://github.com/Debasishray19/vocaltube-speech-synthesis/tree/master/version03

Prof L. Chrostowski Elected to Royal Society of Canada’s College of New Scholars, Artists and Scientists

UBC Electrical and Computer Engineering professor Lukas Chrostowski has been named a Member of the Royal Society of Canada’s College of New Scholars, Artists and Scientists, Canada’s “first national system of multidisciplinary recognition for the emerging generation of Canadian intellectual leadership.”

Recognized by the Royal Society of Canada for “his leadership in research and education in the design of silicon photonic devices and systems for applications in optical communications and biosensors,”

Chrostowski, who has taught silicon photonics workshops and courses since 2008, currently directs the NSERC-funded Silicon Electronic-Photonic Integrated Circuits training program for undergraduate and graduate students and postdoctoral fellows across Canada. Chrostowski holds a BEng in electrical engineering from McGill University and a PhD in electrical engineering and computer science from the University of California at Berkeley. Co-director of the UBC AMPEL Nanofabrication Facility from 2008 to 2017, he is the recipient of a Killam Teaching Prize, has served as an elected member of the Board of Governors and as the Associate Vice President of Education at the IEEE Photonics Society, and co-authored the book Silicon Photonics Design, published by Cambridge University Press in 2015.

Established as an entity of the Royal Society of Canada in 2014, the College of New Scholars, Artists and Scientists identifies and brings together mid-career scholars, artists and scientists who have achieved excellence in their disciplines. It also aims to advance understanding and improve society by promoting “the interaction of diverse intellectual, cultural and social perspectives.”

Chrostowski and the other new members of the College will be inducted this November, during the Royal Society of Canada’s Celebration of Excellence and Engagement in Ottawa.

Chrostowski’s full citation from the Royal Society of Canada reads as follows: Lukas Chrostowski, a Principal Investigator of the Stewart Blusson Quantum Matter Institute (UBC), is recognized for his leadership in research and education in the design of silicon photonic devices and systems for applications in optical communications and biosensors. His present work is focused on developing new photonics-based information processing circuits: neuromorphic processors, quantum communication, and quantum computers.

For more information about the Royal Society of Canada, the newly elected Fellows and the incoming class of the College of New Scholars, Artists and Scientists, please see the UBC announcement and the Royal Society of Canada’s release.

UBC Researchers Find Ways to Hackproof Smart Meters

For the original UBC News media release, please visit: https://news.ubc.ca/2019/06/06/ubc-researchers-find-ways-to-hackproof-smart-meters/

Smart electricity meters are useful because they allow energy utilities to efficiently track energy use and allocate energy production. But because they’re connected to a grid, they can also serve as back doors for malicious hackers. In this Q&A, cybersecurity researcher Karthik Pattabiraman, an associate professor of electrical and computer engineering at UBC, talks about his recent breakthrough aimed at improving the security of these devices and boosting security in the smart grid.

Why is it so important to secure smart meters?

Smart meters are critical components of the smart grid, sometimes called the Internet of Things, with more than 588 million units projected to be installed worldwide by 2022.

In a single household you can have multiple smart devices connected to electricity through a smart meter. If someone took over that meter, they could deactivate your alarm system, see how much energy you’re using, or rack up your bill. In 2009, to cite one real-life example, a massive hack of smart meters in Puerto Rico led to widespread power thefts and numerous fraudulent bills.

Hacked meters can even cause house fires and explosions or even a widespread blackout. Unlike remote servers, smart meters can be relatively easily accessed by attackers, so each smart meter must be quite hackproof and resilient in the field.

How did you approach this problem?

Smart meters are vulnerable to what we call software-interference attacks, where the attacker physically accesses the meter and modifies its communication interfaces or reboots it. As a result, the meter is unable to send data to the grid, or it keeps sending data when it shouldn’t or performs other actions it wouldn’t normally do.

My PhD student and I developed an automated program that uses two detection methods for these types of attacks. First, we created a virtual model of the smart meter and represented how attacks can be carried out against it. This is what we call design-level analysis. Second, we performed code-level analysis. That means probing the smart meter’s code for vulnerabilities, launching a variety of attacks on these vulnerabilities.

Although both techniques successfully discovered attacks against the system, code-level analysis was both more efficient and more accurate than design-level analysis. Code-level analysis found nine different types of attacks within an hour, while design-level analysis found only three.

All of the attacks can be carried out by an attacker with relatively low cost-equipment purchased for less than $50 online, and do not require specialized expertise.

How do your findings improve smart meter security?

Vendors can use the findings to test their designs before they are manufactured, so they can build in security from the get-go. This can make smart meters much harder to crack. By using both approaches—design-level and code-level—you can guard against software tampering on two different fronts.

Our findings can be applied to other kinds of devices connected to a smart grid as well, and that’s important because our homes and offices are increasingly more interconnected through our devices.

Like all security techniques, there is no such thing as 100 per cent protection. Security is a cat-and-mouse game between the attacker and the defender, and our goal is to make it more difficult to launch the attacks. I believe the fact that our techniques were able to find not just one or two vulnerabilities, but a whole series of them, makes them a great starting point for defending against attacks.

The next generation of bionic devices

THE NEXT GENERATION OF BIONIC DEVICES

Inspired by the iconic scene of Luke Skywalker receiving an artificial hand in 1980’s The Empire Strikes Back, UBC is focusing on the future of bionics and the next generation of bio-integrated devices. Led by electrical and computer engineering professor Dr. John Madden, the newly-formed Bionics Network connects over two dozen principal investigators at UBC, Simon Fraser University and the University of Victoria, working in areas such as advanced prosthetics, rehabilitation robotics, regenerative medicine and 3D printed electronics. With specialties ranging across science, engineering and medicine, what brings these investigators together is a desire to deliver on the promise of bionics: the restoration or enhancement of human capabilities.

[Bionic: using artificial materials and methods to produce activity or movement in a person or animal]

BIONICS HEATING UP

2016 was a watershed year for bionics. July saw the foundation of Elon Musk’s new company, Neuralink, which coincided with the launch of the US Defense Advanced Research Projects Agency (DARPA) ’s new Neural Engineering System Design program, both aiming to develop implantable neural interfaces. In October, the world’s first Bionics Cybathlon was hosted in Switzerland, with all the latest prosthetic technology on display. And climactically, in December, after nearly eight years of development, DARPA launched commercialization of its long-awaited, and optimistically-named, LUKE Arm. LUKE (Life Under Kinetic Evolution) represents the cutting-edge of commercial prosthetics: controlled by wireless sensors attached to a patient’s feet, the artificial limb has an anthropomorphic design and natural weight (around eight pounds), 18 degrees of freedom and a responsiveness of one to two seconds using electrical actuators.

Watching the LUKE Arm in motion is an inspiration to many. As one recipient, retired US Army Captain Artie McAuley, put it, “I’m getting to do a lot of things…I never thought I could. This has given me the opportunity to enrich my life.” And more changes are on the horizon. Brain-controlled neuroprosthetics are passing lab tests with flying colours, helping patients with locked-in syndrome to move objects using only their thoughts. Osseo-integration is also gaining traction, which provides more natural and greater control by attaching a prosthetic device directly to the patient’s bone. With this recent wave of new technologies, it may be tempting to believe we’re tantalizingly close to the spectacle of science fiction. But, as Madden puts it, “We’re still a long way from the technological superheroes we see on-screen.”

CHALLENGES TO OVERCOME

This year marks 25 years since the world’s first electrical limb, the Edinburgh Modular Arm System, was unveiled by David Gow. In that time, major advances have seen the production of softer, lighter, stronger and more life-like artificial limbs, leading to the LUKE Arm. However, the challenge of translating lab technology into better patient care is still as great as ever. Studies show that despite these major advances, up to 56 per cent of patients abandon the most cutting-edge electrical devices, opting instead to use simple mechanical systems, which they feel are more reliable and less complicated.

The complexity of the human arm and the challenge it represents is a true marvel. Not only does it have great agility, with 34 degrees of freedom (27 in the hand, seven in the arm) and the ability to manipulate a wide range of differently-shaped objects, but it also has fast twitch response (hundreds of milliseconds) and an average grip strength of around 100 pounds. In addition, the human hand contains over 17,000 nerves capable of detecting weight, firmness, shape, temperature and texture, feeding this information back to the brain via a two-way signalling network. And, if that wasn’t enough, the whole system is wrapped in a soft exterior suitable for human-to-human interaction.

HOW CAN BIONICS DELIVER ON ITS PROMISE?

The LUKE Arm would be indistinguishable from magic 50 years ago and it stands as a testament to how far we’ve come from Götz von Berlichingen’s sixteenth century mechanical hand. But there is much ground to cover in bringing artificial devices to life. “Translating our technologies into better patient care is the goal of the Bionics Network” says Madden. “This means working across boundaries to integrate teams and develop a new generation of devices that are softer, smarter and more sensitive.” For the Bionics Network, this will involve three key research themes: 3D printing to produce custom-fitted parts and reduce costs; soft robotics to exchange rigid components for softer, smarter materials; and bio-integration, to create devices that are compatible with the human body. Madden’s team at the Molecular Mechatronics Lab recently hit the headlines with the development of a soft artificial skin that can impart touch sensation to robots.*

At a time when bionics is more popular than ever, the Bionics Network is developing a network of scientists, engineers and clinicians to be at the forefront of this new generation of devices. By combining human biology, biomimetic design and the perspective of the patient, the group aims to deliver mobility, freedom and life back to many.

Bionics Network is supported by UBC’s Vice President of Research and Innovation Office, in collaboration with the Institute for Computing, Information and Cognitive Systems (ICICS).

Original article on UBC Applied Science

Work with a Purpose: AbCellera Hires Talented People who want to make a Difference

Article by Michael Smith Labs Communications Team

You might already hold the cure for flu or dementia. How? Your body makes billions of unique antibodies and one of these could be the basis for a treatment. But, how do you find that one rare therapeutic antibody buried in the proverbial haystack?

AbCellera, a local biotechnology company founded at the University of British Columbia (UBC), has developed a new method that can search immune responses more deeply than any other technology. Using a microfluidic technology developed at UBC’s Michael Smith Laboratories, advanced immunology, protein chemistry, performance computing, and machine learning, AbCellera is changing the game for antibody therapeutics.

In 2012, AbCellera started with only six members. The company has grown to over 75 employees who contribute expertise in diverse fields such as computational and data science, cell biology, protein chemistry, and biophysical engineering. This interdisciplinary team is crucial to the discovery of each antibody screened at AbCellera.

Protein scientists prepare the production of antigens to screen for antibodies with the right binding properties. These antibodies are selected from millions of antibody-producing cells using microfluidics, biochemical assays, robotics, and machine learning. Afterwards, next-generation sequencing is used to read the genes that code for the antibodies. Making sense of the huge datasets that are produced from millions of cells requires high-speed computation, bioinformatics, and some next-level visualization of the resulting datasets. Out of millions of sequences, only a couple dozen of the best will be expressed and evaluated for their ability to become a viable treatment.

From start to finish, the process of antibody discovery can be done in five days, compared to the three months required for conventional techniques. This technology could drastically speed up the drug discovery process for AbCellera’s impressive list of partners and for patients suffering while they wait for a cure.

According to Dr. Carl Hansen, Director and CEO of AbCellera, the company has a rather unconventional hiring philosophy. In addition to drawing talent from obvious programs like biotechnology, AbCellera recruits from electrical engineering, computer science, finance, and management. The company’s broad recruiting strategy has attracted some of the best talent at UBC.

“In British Columbia, we have world-class science and expertise, particularly in genome science, stem cell research, engineering, and computation…[but] there’s not a huge biotech sector in Vancouver,” says Hansen. “So, if you were to take a classic hiring approach and hire people who have only had years of experience in pharma and biotech, you’d be overlooking a lion’s share of the great talent we have locally.”

Hansen continues by describing the ideal candidate as someone “who has passion, can learn quickly, and is interested in being a part of something that’s growing and at the cutting-edge. Someone who wants to make an impact.”

Not only is AbCellera impacting the way antibody therapies are developed, but its booming business is creating jobs in British Columbia’s up-and-coming biotechnology sector. Recent biological breakthroughs, increasing application of powerful computation, and an influx of capital, have created a fertile ecosystem for world-changing discoveries. It will be exciting to see what the future holds for biotechnology in British Columbia, and for rapidly growing companies like AbCellera.

Robotics and Control Lab wins student paper awards at SPIE Medical Imaging 2019

Congratulations to the Robotics and Control Laboratory, whose team has received both the first place and the second place student paper awards in image guided procedures, robotics Interventions and modeling at SPIE Medical Imaging 2019.

The first place paper is titled “EpiGuide 2D: visibility assessment of a novel multi-channel out-of-plane needle guide for 2D point of care ultrasound”.

The second place paper is titled “Temporal enhanced ultrasound and shear wave elastography for tissue classification in cancer interventions: an experimental evaluation”.

The conference was held from February 16 to February 21 in San Diego, California.

Spherical display brings virtual collaboration closer to reality

Virtual reality spherical display. Credit: Clare Kiernan

For more information, contact Lou Corpuz-Bosshart

Virtual reality can often make a user feel isolated from the world, with only computer-generated characters for company. But researchers at the University of British Columbia and University of Saskatchewan think they may have found a way to encourage a more sociable virtual reality.

The researchers have developed a ball-shaped VR display that supports up to two users at a time, using advanced calibration and graphics rendering techniques that produce a complete, distortion-free 3D image even when viewed from multiple angles.

Most spherical VR displays in the market are capable of showing a correct image only from a single viewpoint, said lead researcher Sidney Fels, an electrical and computer engineering professor at UBC.

“When you look at our globe, the 3D illusion is rich and correct from any angle,” explained Fels. “This allows two users to use the display to do some sort of collaborative task or enjoy a multiplayer game, while being in the same space. It’s one of the very first spherical VR systems with this capability.”

The system—which the researchers are calling Crystal—includes a ball-shaped display with a diameter of 24 inches (600 millimetres). The hollow spherical display surface was custom-made to specifications in Ottawa, while four high-speed projectors and one camera used for creating the images, calibration and touch sensing were purchased off-the-shelf.

The researchers are working on a four-person system and see many potential uses for their display in the future, including multiplayer virtual reality games, virtual surgery and VR-aided learning. However, they are focusing on teleconferencing applications and computer-aided design for now.

“Imagine a remote user joining a meeting of local users. At either location you can have a Crystal globe, which is great for seeing people’s heads and faces in 3D,” said Ian Stavness, a computer science professor at the University of Saskatchewan and a member of the research team. “Or you can have a team of industrial designers in a room, perfecting a design with the help of VR and motion tracking technology.”

While the technology is young, the researchers are forecasting a good future for it.

“We’re not saying that spherical VR will replace flat screens or headsets,” said Fels, adding “but we think it can be a good option for VR activities where you still want to see and talk to other people—be it at home or in the office, for work or play.”

The research was supported by the Natural Sciences and Engineering Research Council of Canada and B-Con Engineering.

Original article on UBC News

Karthik Pattabiraman and Matei Ripeanu receives 2018 Faculty Research Awards

Matei Ripeanu (left) and Karthik Pattabiraman (right)

ECE professors Karthik Pattabiraman and Matei Ripeanu are recipients of UBC’s 2018 Faculty Research Awards. Professor Pattabiraman received the UBC Killam Research Prize, which recognizes outstanding research and scholarly contributions. Professor Ripeanu received UBC’s Killam Research Fellowship, which enables faculty to pursue full-time research during a recognized study leave.

Professor Pattabiraman’s work in the area of dependable computer systems has led to the adoption of low-cost techniques to protect everyday common computer systems, such as desktop computers, mobile phones and self-driving cars, from errors and failures. He has also received numerous prestigious accolades and awards, most recently a Distinguished Alumni Educator Award for Early Career Educator from the University of Illinois at Urbana-Champaign (UIUC). Professor Pattabiraman’s transformative research has significantly impacted academia and industry, and continues to shape the future of dependable computer systems. For more details on professor Pattabiraman’s research, please see his website: blogs.ubc.ca/karthik

On his upcoming study leave professor Ripeanu will focus on two projects: solutions to process time-evolving large-scale graphs (with collaborators at Lawrence Livermore National Laboratory) and blockchain-based solutions for markets in computing resources (with collaborators at iExec). Professor Ripeanu was recently recognized for his impact as a mentor having received the 2018 Killam Award for Excellence in Mentoring. For more details on professor Ripeanu’s research, please visit his group website: netsyslab.ece.ubc.ca

Full list of this year’s Faculty Research Award winners is available here.

The future will be faster thanks to photonics

Hassam Shoman holding the chip developed by his research team. (Photo: Erica Yeung)

Imagine standing in a checkout line at a store with 10 other people, waiting for one cashier to serve you. Now imagine how much faster it would be if there were 10 cashiers available. Similarly, the operation of a silicon electrical chip can be substantially improved by introducing photonics: more information is routed using photonics wires on chip allowing information to be processed more efficiently compared to traditional copper electrical wires.

“Instead of serving one customer at a time, you’re serving 10 customers at the same time,” explains Hossam Shoman, a PhD student in the Department of Electrical and Computer Engineering. “The multiplexing of frequencies that make up light makes it faster because you’re sending more information at once.”

Hossam is part of a research team, which includes ECE graduate Hasitha Jayatilleka and professors Lukas Chrostowski and Sudip Shekhar, that recently designed an integrated element that can monitor light precisely and control it on a chip using just a single element. This means better on-chip real estate management in devices, resulting in smaller devices without sacrificing performance.

“This will reduce the footprint compared to other devices and allow for large scaling of photonic systems,” says Hossam. “You can continue boosting the performance of devices by increasing the number of cores of devices, which is becoming difficult to do using electrical wires because they overheat and decrease the performance of the chips.”

Hossam believes that their research will benefit the biomedical industries by enabling the development of cost-effective biosensors that can help detect early stages of cancers. The technology will also make medical devices more accessible and affordable.

“We’re working on integrating and packaging the chips in cost-effective ways for consumers and countries that can’t afford expensive medical equipment,” says Hossam. “Imagine a sensor on the back of your phone that could detect your blood sugar level.”

The paper, titled “Photoconductive Heaters Enable Control of Large-Scale Silicon Photonic Ring Resonator Circuits,” was recently published in Optica.

Photo gallery.

Professor Matei Ripeanu receives Killam Award for Excellence in Mentoring

Photo credit: Lukas Chrostowski

Electrical and Computer Engineering (ECE) professor Matei Ripeanu is the recipient of the 2018 Killam Award for Excellence in Mentoring in the mid-career category. The award recognizes outstanding mentorship of numerous graduate students over many years.

Professor Ripeanu’s approach to mentoring is very much catered to the needs of his students.

“I try to match students with projects that best fit their skillset and discover with them specific problems they care about so that they can find their maximal motivation,” says professor Ripeanu.

His students will be the first in line to acknowledge the impact of his mentorship and why he is an excellent advisor and mentor.

“One aspect that stands out concerning professor Ripeanu’s mentoring approach is his adaptability,” explains Hassan Halawa, a PhD student in the department. “He deliberately adjusts it to each mentee – according to their particular traits, skill sets, motivations, circumstances and ambitions.”

“His mentoring style is to let the students’ curiosity and interest be the driver of their own research journey,” says Hao Yang, an ECE masters student.

There are four aspects to professor Ripeanu’s approach to mentorship:

Find out in what areas students can excel
Discover what projects motivate them
Fostering a climate of respect and intellectual honesty, and
Enabling students to own their career path and make sure they do not lose their sense of happiness in the process.

Ultimately, professor Ripeanu hopes that his students will find success both professionally and personally.

“My experience is that if we are successful with the above, then the desired professional and often personal accomplishments follow.”

Learn more about professor Ripeanu’s research at Networked Systems Laboratory.