It’s exciting to be at the forefront of a movement, a societal shift that changes the way our lives look, feel, and sound.
Audio technology is such a movement. The authors of a recent article examining the future of audio technology state, “Just as the Internet, mobile, and cloud gave rise to new innovations across consumer and enterprise [spaces], we believe audio [technology] will fundamentally change how people connect and businesses operate”(1).
We use increasingly-ubiquitous smart speakers in our homes, asking them to play our favorite songs and set timers as we cook and bake (crucial given the stuck-at-home baking trends). It seems the pandemic has amplified the progress and popularity in this medium of sound.
Fortunately, we just happen to have a few sound experts at PredictionHealth and we’re getting their thoughts on all things audio today!
Dan Gurney is Lead Engineer at PredictionHealth, but he’s also our accordion-playing, folk music aficionado. He even built a company specializing in streaming online concerts.
Dan, tell us how music fits into what’s going on in sound technology today, and how that relates to PredictionHealth.
Home recording via a laptop or desktop computer has made it much easier to record music, and podcasts have further intensified that trend. Now the recording capabilities in computing devices are getting extremely good, and luckily we can utilize these for physician conversations in PredictionHealth.
Have there been any moments, or specific products, which were game changers in making music accessible?
Garageband on Macs brought a ton of people into the home recording world. I love the idea of giving creative tools to people and seeing what they come up with. Nowadays, as computers have gotten faster, smaller, and cheaper, so have home recording setups. Now people can rival professional studios with some relatively inexpensive gear… but acoustics still matter. I imagine AI will be the next big revolution in all of this. You can already see it with decent phone cameras that use AI to enhance photos.
What is it about music that never gets old to us (I mean, there are ALWAYS new and great songs being written), and does that translate to certain audio technologies being around for the long haul?
Music is something that seems to universally connect with people across language barriers. It must be hardwired into us at some basic human level. Given that so many people want to listen to music digitally, that has led to the creation of audio formats like mp3 and wav (which we use here at PredictionHealth). Wav is lossless and used in all kinds of applications, and as such it has been around for years and years now.
Do you think music and coding have certain connections and that’s why you love both?
Absolutely. They both seem to occupy a part of my brain that uses creativity and logic at the same time. Writing a good piece of code feels a bit like playing music and in the best of situations, gets me into the flow state. Bad code and bad music…vice versa. A code editor and an instrument, to me, are mediums for expressing myself. One time a startup advisor meant to insult me by calling me “an artist” but I thought that was very nice of him!
Dan really is an artist. Check out his music!
Daniel Weidman has been researching varied aspects of speech and sound for years before starting as an Engineer at PredictionHealth. He’s explored such detailed topics as hearing speech in noisy environments and whether auditory feedback can help improve gait for patients with multiple sclerosis.
Daniel, What is audio? How does it work in humans?
At its most basic level, audio consists of tiny vibrations through the air that are sensed by the tiny hairs inside of our ears and then transformed by our brain to facilitate what is perhaps our most important sensory input. Our brains employ all sorts of tricky processing techniques to extract speech signals from noise—techniques that the scientific community is just beginning to understand.
What about people who can’t hear and where does technology stand in terms of increasing accessibility for the hearing-impaired?
A lot of the modern technology for people with hearing loss has centered on signal processing. While the concept of simply amplifying audio to make it easier to hear is nothing new—ear trumpets have been a thing since at least the 1600s, the practice of selectively deciding what parts of a sound signal to amplify—and how those parts can be chosen in a manner customized for a particular listener—is newer.
Companies have used AI to individualize programming for cochlear implants and predict user outcomes.Advancements in speech recognition have brought far better automated transcriptions and captioning. There are now plenty of smartphone apps that can provide serviceable audio transcriptions to persons with hearing impairment in real-world situations. Pedro even experienced this with his own patients during his medical training.
What are all of the ways that sound determines what we do, think, and feel?
The sense of hearing is completely intertwined with the rest of the body and its senses. For instance, there is evidence that auditory ability is linked to balance and risk of falls in older adults. Hearing aids may even improve the balance of a person with hearing impairment, perhaps by helping free cognitive resources from focusing on difficult-to-hear noises, or by enabling a person to pick up on tiny environmental cues that help them understand the events going on around them.
Likewise, the senses of sound and sight are closely linked. One’s ability to interpret a speech signal when they are able to see the face of the talker is usually significantly better than when they cannot—particularly in the case of a sighted person with hearing impairment.
The way that the brain can take these two, completely different types of sensory inputs and combine them in order to better understand the situation in real time is something that may become increasingly significant in the realm of artificial intelligence and, in particular, speech recognition (2).
Sound patient care (last pun, we promise)
We’ve evolved over millions of years to use sound as part of understanding our environment and our environment and our social interactions. Hearing movement around us, wind in the trees, but also the voices of our friends and family – what they say and how they say it – imbuing statements of fact with emotion that colors things in a complex and beautiful way. We believe this applies just as much to patient-clinician conversations. At PredictionHealth, we’re documenting the facts to save our clinicians hours per day, but we’re just as enthusiastic about building systems that can understand how things are said. Just as the experienced clinician uses subtle patient cues to better understand and make clinical decisions, we can use AI to perceive the deeper meanings in conversation. To continually improve patient care, we’re building all these smarts right into the systems clinicians use every day to take care of people.
1. The State of Sound in 2020 and Beyond | BVP
2. Deep Audio-Visual Speech Recognition | arXiv
Get in touch with us to learn more about using PredictionHealth in your clinic. We’d love to hear from you.
Dan Gurney and Daniel Weidman