New Audio and Sensor Features Highlight the Importance of Edge Processing
November 14, 2019 - Mike Adell and Adam Abed
Leading mobile, ear and IoT device makers are constantly looking for differentiation that will drive end-user adoption and sales. For example, over the past few years, camera quality has been a primary driver for smartphone differentiation as consumers have been demanding high quality photos and videos for personal and social media postings. These cameras, with multiple usage lenses, high megapixel capture, and intelligent software can make almost any picture look amazing. With these technological developments, the phone cameras are now reaching a quality standard to satisfy most consumers. Moving forward, end-users and OEMs will be looking for the next “big” thing to improve their daily life experiences. What is that next big thing? One possibility will be turning your handset or smart device into an intelligent yet private personal assistant.
So how do we transform these products into intelligent personal assistants? There are two primary areas of focus on. First, the device needs to have access to the right information in order to help you with various tasks in your daily life, e.g. giving your device senses. This requires enabling abilities like hearing (microphones), seeing (cameras and light sensors), speaking (speakers), location and connection to other devices (GNNS, WiFi, Bluetooth). Second, your device needs to have the real-time, intelligent computing but without draining your battery life or invading your privacy. How do we achieve this? The answer is to leverage a low power, secure, custom edge AI processor that utilizes machine learning to perform intelligent tasks efficiently, while sending as little information up to the cloud where a user’s privacy can be compromised.
Let’s take a closer look at how intelligent edge multi-sensory processors can help enable your mobile, ear and IoT devices to be better and more secure assistants.
One key example we see today is the “voice assistant” feature on your phone or smart device. Presently we can talk to our device by saying a pre-defined keyword (e.g. “Alexa” or “OK Google”) followed by a task we want to perform (e.g. “what is the weather” or “set my alarm”). Currently edge processors detect the keyword then immediately send the information to the cloud where it is interpreted and acted upon. A big problem is that once your audio data is in the cloud it can be hacked or used by third parties in an undesirable way. How do we keep this cool feature while keeping your information private? The answer is to utilize recent advances of edge AI processors and perform the command interpretation and response logic on the device, locally, “at the edge.” This has several important benefits. First, your sensitive personal audio data stays local, without being sent to the cloud where it can be used against our wishes. Second is the ability to perform tasks significantly faster by processing our requests on the device versus sending that information through a slow cellular or WiFi connection to the cloud and back. Your assistant is now not only much more private but it can respond so much faster, making users interactions much more natural. This is a great example of how edge AI processors can advance existing use cases to maximize the helpfulness of the devices we use and trust every day.
What’s next? Not only can edge AI processors advance existing features but they can also create new unique experiences. By utilizing all of the sensor information around your device, edge AI processors can understand your environment and situational context and then help you naturally without you having to think about it. One common example is the edge processors ability to know whether you are in your car (this can be performed based on audio cues, motion cues, etc.). Once it knows, it can then automatically launch your navigation application and ask you where you need to go - no more fumbling with your car navigation or phone app. When you get to your destination, the edge processor will know you have stopped and can save your parking location so you know exactly where to go upon your return. There are so many examples of useful situations where your smart device knowing your environment or context can make your life easier. Context awareness features like this must be run at the edge for saving power (it is always-sensing so it cannot drain much battery power to perform), for improving response (latency - going to the cloud), and for securing privacy (keeping your information being sent to cloud).
These features are emerging now and will soon become expected (or demanded) by consumers. Machine learning and audio processing done efficiently is the key to enabling such applications without impacting your battery life and making your new personal assistant more effective. The best way to enable this is through edge AI processors.
What do you want your personal assistant to do next? To learn more about audio edge processors, visit www.aisonic.com