Qualcomm presented Snapdragon Sound Elite Gen 2, an audio platform that combines Ultra-low power Wi-Fi 6EBluetooth, sound processing and local artificial intelligence acceleration. It should allow future headphones and wearable devices to receive lossless sound up to 192 kHz and connect directly to cloud services, without systematically using a smartphone.
Wi-Fi complements Bluetooth rather than replacing it
Wireless headphones almost always use Bluetooth, whose consumption is suitable for small batteries but whose speed and range impose limits. Snapdragon Sound Elite Gen 2 integrates Wi-Fi 6E with support for the 2.4, 5, and 6 GHz bands. Qualcomm’s XPAN technology must choose the right network and keep consumption low enough for a compact device.
Manufacturer announces lossless music streaming up to 192kHz over Wi-Fi, compared to more restrictive speeds on Bluetooth. The wider link should also reduce head tracking latency for spatial audio. A user could continue listening in multiple rooms as long as they remain covered by their network, rather than relying on phone proximity.
The direct connection to the cloud is the most ambitious difference. Headphones, audio glasses, or a hearing aid could contact an online service on their own. This principle should be used in particular for voice assistants, translation or other applications that require more calculations than the integrated chip can provide.
This autonomy does not mean that all uses will work without configuration. The device manufacturer will need to integrate the platform, develop associated services, and define how Wi-Fi or cloud identifiers are managed. Qualcomm has not yet announced any commercial headsets, pricing, or consumer availability date.
128 GOPS to process sound and context locally
The platform promises to double AI computing power, with 128 GOPS. It brings together general computing, audio, sensors, connectivity and neural acceleration in a single package. Qualcomm thus focuses on tasks that are carried out continuously: adjusting the sound based on the environment, improving microphones, adaptive transparency, activity detection or listening personalization.
Some of these features are based on Qualcomm’s fifth generation of adaptive active noise reduction. The Kymera architecture distributes operations between several specialized blocks: an ARM M55 core for main processing, two Hi-Fi 5 DSPs for offloading certain tasks, and a Hi-Fi 3 DSP dedicated to low-latency operations.
The component is not limited to traditional headphones. Qualcomm cites audio glasses, headphones, open products that do not block the ear, and future multi-modal accessories. We just saw Meta’s new cameraless Audio glasses, a format that will likely benefit from a platform that combines voice, sensors and networking.
A more integrated chip, but one that promises to be tested
Qualcomm announces reduction of 30% of the thickness of the platform compared to the previous generation. This gain can leave more room for the battery or help manufacturers design thinner temples for glasses and headphones. The company also highlights improved energy efficiency, without providing a standardized autonomy measure applicable to a finished product.
This announcement comes two days after the presentation of the Snapdragon 8 Elite Gen 6 for smartphones. The strategy is similar: move more computing to the device while maintaining access to the cloud when load demands it. In audio, however, the result will depend a lot on the size of the battery, the quality of the antennas and the actual frequency of Wi-Fi use.
The platform also supports aptX Lossless and adaptive noise reduction, but Qualcomm has not provided a list of committed manufacturers or an initial release schedule. Therefore, the first products will need to demonstrate that Wi-Fi connection and always-on processing are still compatible with a day of listening on small batteries.
