This podcast was hosted by Maxwell Murray, Radian Podcast, and can be viewed in its entirety here.
The idea for Canary Speech started in a bagel shop, where two longtime friends—one a former Amazon AI leader, the other a neuroscience researcher—asked a deceptively simple question: What if voice could be used to detect disease? They weren’t talking about what people say, but how they say it: pitch, tremor, tempo, and thousands of other micro-markers buried in everyday conversation. That moment launched a company built around a big idea: voice is the most natural, information-rich signal we have, and it’s vastly underutilized in healthcare.
Today, Canary’s technology listens in real time, analyzing speech in ten-millisecond intervals, segmenting speakers, and flagging patterns consistent with depression, anxiety, cognitive decline, and more. Integrated through a simple API, it runs passively in the background of clinical conversations and telehealth visits, delivering insight without disruption. It’s not replacing doctors but rather equipping them with more context when it matters most.
This goes beyond NLP and sentiment analysis. Language is shaped by culture, age, and education; but the mechanics of speech, our vocal biomarkers, are deeply physiological. That makes Canary uniquely positioned to uncover changes that traditional tools might miss. It’s ambient, fast, and invisible to the patient. And in a world where clinicians are being asked to do more with less, that kind of superpower can make all the difference.
