Signal Processing Pioneer’s Tech Has Improved Autism Diagnosis

Shrikanth Narayanan has spent his entire career making speech and language processing technologies more accessible.

The IEEE Fellow has developed machine intelligence and signal processing technologies to analyze human behavior including spoken language, facial expressions, and physiological indicators.

Thanks to his work, medical professionals can better diagnose and monitor autism, depression, and other conditions.

Anyone using digital assistants has benefitted from Narayanan’s research in understanding and interpreting human emotions from speech. The assistants are now more intuitive, and they can better understand and respond to a user’s commands.

It’s also easier now to learn a new language thanks to tools he developed that provide feedback on how to pronounce words.

Narayanan is a professor of electrical engineering, computer science, linguistics, psychology, pediatrics, and otolaryngology at the University of Southern California, in Los Angeles. He also heads USC’s Signal Analysis and Interpretation Laboratory and holds several other academic positions within the university. He is a visiting faculty researcher at Google DeepMind in Los Angeles.

Narayanan received the 2025 IEEE James L. Flanagan Speech and Audio Processing Award for his contributions to speech communication science and technologies for inclusive human-centered engineering. The award is sponsored by Mitsubishi Electric Research Laboratories.

“I am so touched and honored,” he says about getting the award. “I started my career at Bell Labs, and James Flanagan was a legendary researcher in speech and audio there. Many people who have received this award have been my heroes in the field—who I look up to. Their work has inspired me profoundly.”

An early fascination with how the human body functions

Growing up in Chennai, India, Narayanan wanted to be a physician because he was fascinated with how the body works. He applied and was accepted into medical school at the age of 17, but his career plans changed before he even stepped into a classroom.

Narayanan’s father was a chemist, and his uncle was an electrical engineer. After several discussions, his family persuaded him to switch to engineering even at the “supportive protest” of his uncle who was an engineer, he says.

“At the time, electrical engineering was touted as the most foundational field of science,” he says. “I didn’t know much about it, but it soon became clear to me that I could start matching how signal processing systems work to conceptualize how the human body functions. That made me this sort of engineer who is very human-focused right from the beginning. I look at people from an engineering angle.”

He earned a bachelor’s degree in EE in 1988 from the College of Engineering, Guindy, (now part of Anna University, in Chennai). Narayanan went on to earn his master’s and doctoral degrees in EE in 1990 and 1995, from the University of California, Los Angeles.

He started his career as a research scientist in 1995 at AT&T Bell Labs (now Nokia Bell Labs) in Murray Hill, N.J. While working on speech and language processing technologies, he noticed that the applications being developed were only for healthy adults, so he and other researchers decided to focus on ones for children.

“When we started working on technologies for children, we immediately found fundamental challenges because of this dynamic trajectory of how their speech and language changes,” he explains. “As children are growing, they’re developing not only physically and physiologically but also socially.”

The researchers first had to create a foundation based on speech science for the changes to be studied objectively and quantitatively, he says.

“Speech and language result from a complex orchestration of various processes that happen in the brain and the neural and motoric systems,” he says.

“My greatest joy is working with my students in my lab and learning from them, more than being a teacher or advisor. It’s amazing that I get to learn new things every day.”

To study the processes in a systematic way, the researchers used sensors and imaging to measure changes in speech and language skills. After collecting data in the form of signals, the researchers applied signal processing techniques to extract meaningful information.

Narayanan concluded that their method could be used for children who have developmental conditions such as autism spectrum disorder, language delays, and similar disorders.

They invented behavioral signal processing (BSP) technology, which analyzes and interprets speech and language in social situations. Narayanan says the technology is useful for children with autism who typically have a difficult time with social interactions. The researchers also developed computational models to detect and interpret emotional cues from autistic children’s speech and facial expressions.

Another tool they created monitors the progress of the communication abilities of children who are not developing language skills at the expected age.

The researchers’ early work in understanding and interpreting human emotions from speech has inspired features used in virtual assistants such as Alexa and Siri to sound more natural and recognize a user’s emotions. BSP technology helps the devices recognize not only what users say but also how they say it.

The researchers’ work in acoustic modeling, language modeling, and integrating contextual information enabled digital assistants to identify speech more accurately.

Tech to improve mental health

Narayanan left Bell Labs in 2000 to join the USC faculty. He always wanted to mentor students and work with people from different disciplines, he says, so when he was offered a teaching position in California—a place he loves—he decided to give it a shot.

“My greatest joy is working with my students in my lab and learning from them, more than being a teacher or advisor,” he says. “It’s amazing that I get to learn new things every day.”

Throughout his nearly 25 years at USC, Narayanan has continued to develop speech and language processing applications for health care. He uses technologies such as BSP to create methods to better understand mental health.

“Bringing engineering tools to support research into mental health has been a big area,” he says. “I’m very committed to that field.”

Diagnosing and treating mental health conditions often involves interacting with patients using speech and language. In psychotherapy, for example, a mental health professional talks with the patient to identify troubling thoughts, emotions, and behaviors and to help address them.

Psychotherapy research and clinical practice tend to use manual methods to collect and evaluate performance and efficacy data, Narayanan says, but that is not scalable and can lead to inaccuracy. The answers might not actually reflect how the patient feels, he says.

Narayanan and his colleagues invented a way to collect data through speech and language-based biomarkers to characterize therapy quality and outcomes. They also designed objective measures to detect and monitor a person’s speech patterns for signs of depression and anxiety.

He currently is working with the U.S. Defense Advanced Research Projects Agency to identify biomarkers for people with suicidal ideation.

Narayanan holds 19 U.S. patents and has helped to found several startups to commercialize his technologies.

Overseeing USC’s grand research plans

In February he took on a new role that utilizes his multidisciplinary background: USC appointed him as vice president for its presidential initiatives, a newly created position. He coordinates and expands the reach of the university’s research initiatives in computing, health, and sustainability, things the university refers to as moon shots. Notably the university has invested more than US $1 billion in its Frontiers of Computing initiative.

“The university and its president have this big strategic vision of thinking about grand problems, like the future of health, the future of computing, and sustainability of the planet,” Narayanan says. “They wanted a researcher and a scholar who works across disciplines. They want me to connect people and ideas to launch these big initiatives that have a global footprint.”

“Advances are taking place at an astonishing rate in the evolving fields encompassed by our moon shots,” Carol Folt, the university’s president, said in an announcement about the appointment. “This role was created to focus not only on implementing but also continually broadening, amplifying, and weaving our moon shots together so USC remains at the forefront of discovery and innovation. Professor Narayanan is the perfect choice for this role.”

IEEE: A big family

At the encouragement of one of his undergraduate professors, Narayanan joined IEEE in his senior year.

“I realized IEEE is a home to learn, to share, and to constantly grow,” Narayanan says. “IEEE provides that for us. It’s a platform to situate your work in your field, and in the broader context of society and humanity. And, of course, you make a lot of lifelong friends, and you give back as a volunteer.”

And give back he has. A member of the IEEE Computer and IEEE Signal Processing societies, he was the latter’s first vice president of education.

He has been on the editorial boards of both societies’ publications and has served as editor in chief for their journals and transactions. He also held leadership roles in organizing the societies’ conferences and workshops.

Both societies have recognized him for his work. He received an IEEE Computer Society McCluskey Technical Achievement Award this year and an IEEE Signal Processing Society Shannon-Nyquist Technical Achievement Award last year.

Volunteering has become part of his life, he says, and over the years, he has encouraged his students to join.

“Many of them are now professors around the world, and they encourage their students to join,” he says. “IEEE is like a big family.”

This article appears in the March 2025 print issue as “Shrikanth Narayanan has Improved Autism Diagnosis.”