“What AI, and specifically DNN, does is a data-based approach for the device to become smarter and deal with challenges that are thrown at it from an acoustic viewpoint,” stated Achin Bhowmik, the CTO of Starkey. This approach is eventually making its way into other wearables such as smart glasses, including the recent update from Meta in the Ray-Ban Meta and Oakley Meta HSTN smart glasses.
The newly introduced **Conversation Focus** functionality is a direct solution to one of the biggest sources of irritation associated with public areas: the ability to focus one’s attention on a specific conversation surrounded by a cacophony of sounds. Utilizing the beam-forming microphone and open-ear speakers, the device has the ability to focus the voice of the person standing right in front of the wearer above the rest of the sounds by applying some form of AI machine learning-related signal processing, similar to those advanced hearing aids employing machine learning algorithms for improving hearing ability by differentiating the target voice and reducing the effect of general voices, thus not completely muffling the surroundings, as done by regular noise-cancelling headphones.
On the other hand, Meta’s execution enables sophisticated control. This includes swipe control on the right part of the frames to manage amplification or modify it through adjustments in device settings to adapt to all kinds of situations, including restaurant or train scenes with many people. Hands-free starting through voice commands like “Hey Meta, start Conversation Focus” ease use. Together with tap-and-hold buttons, this allows quick manual starting. Initially, users in America and Canada will get this update through Meta’s Early Access.
Where the engineering is, is in trying to successfully integrate the processing power needed to do computations in real time to achieve audio isolation with the constraint of battery life. Although hearing aids are available today that use specific hardware to run DNNs in real time, Meta’s technology has to factor in not only audio but also camera, and various other AI-related computations. The open speaker also does not allow any panning in the acoustic path.
The second significant addition that came along with this update is a multi-modal AI integration service with Spotify. The glasses come equipped with a camera and computer vision, allowing it to detect some visual cues, an album, or even some festive items, and automatically play some relevant music. The voice command tool, “Hey Meta, play a song to match this view” initiates this feature that enables a combination of vision detection and a Spotify recommendation system that delivers contextually driven playlists. The convergence of vision detection and personalized audio playlists is a forecasted direction that AR wearables might take, where vision detection informs AR digital responses.
Technologically, this involves rapid image classification, either on or in the cloud, correlating identified objects with metadata that Spotify would be able to employ in their playlists. The smart glasses’ AI system must be able to cope with various lighting, partial occlusion, and conceptual visual themes with minimal delays. This music matching functionality, although now more for show, shows the potential that smart glasses have in connecting various inputs, such as sight and sound, into an integrated and dynamic experience.
The hardware in Meta’s second-generation Ray-Ban glasses has these capabilities integrated into it, along with longer battery life, measuring eight hours, as well as higher camera resolution. More importantly, Meta is not planning on implementing these software improvements only on their newer models, as these improvements will also extend to those who possess first-generation models, which highlights that the company prioritizes software capabilities via AI over hardware improvements.
In the context of wearable technology, the above-mentioned advancements position Meta’s glasses amidst a growing category of consumer electronics that eliminate the divide between assistive technology and lifestyle products. Apple’s AirPods Pro Conversation Boost or the clinical-grade hearing aid mode and Starkey’s Edge Mode+ for hearing aid users anticipate this trend with the rising use of AI-powered audio personalization for mainstream consumers. This stems from the increasing capabilities of algorithms to analyze the context rapidly, possibly closing the divide between hearing assistance and entertainment delivery or a single approach to context-driven audio optimization.
For early Adopters, the pairing of useful hearing aids and fun multimedia use is a strong example of the potential for AI to make their lives better. Helping to distinguish a buddy’s voice in a busy pub or initiating a precisely timed soundtrack based on a vacation memory, Meta’s Glasses demonstrate that wearable AI has a bright future ahead of it, where the goal becomes not simply to respond, but to adapt.
