Qualcomm’s Snapdragon Sound Elite Gen 2 Brings Smartphone-Free AI to Headphones

MAUI, Hawaii — Qualcomm has unveiled the Snapdragon Sound Elite Gen 2, an AI-focused audio chip designed to let headphones, earbuds, smart glasses and other wearable devices run AI assistants without maintaining a connection to a smartphone.

Announced at Qualcomm’s annual Snapdragon Summit, held in Maui from September 21–24, the chip can support on-device apps, audio processing and connections to cloud-based services such as ChatGPT when Wi-Fi is available. It could enable wearable devices to collect audio and video data from their surroundings and use that information to provide more context-aware assistance.

What can Snapdragon Sound Elite Gen 2 do?

Unlike many current wireless earbuds, devices built with the new platform could host apps in their own onboard memory rather than relying continuously on a connected phone. Users could make voice requests and ask an AI assistant to perform tasks such as adding an event to an online calendar or making an online purchase.

Qualcomm hopes the chip will be used in a broad range of “hearables,” including:

• Over-ear headphones
• Open-ear audio devices
• Audio glasses
• Camera buds with built-in cameras
• Hearing aids
• Future personal AI wearables such as pendants and clips

How the AI audio chip works

For users who want to connect a hearable to a smartphone or computer—for example, to operate a visual interface—the platform uses a low-power Wi-Fi channel alongside Bluetooth. Qualcomm says architectural improvements at the chip level significantly reduce Wi-Fi power consumption compared with standard Wi-Fi connections.

The low-power Wi-Fi connection has longer latency and lower peak speeds than conventional Wi-Fi. However, Qualcomm representatives said Wi-Fi can provide better playback quality than Bluetooth for audio devices because Bluetooth compresses data during transmission.

Adaptive noise cancellation and lossless audio

Adaptive and selective active noise cancellation will allow users to filter or emphasize sounds on demand. For example, a listener could hear important sounds such as emergency sirens while walking, or emphasize nearby voices during a conversation while reducing other background noise.

This noise-cancellation technology uses standard machine-learning algorithms rather than the large-scale transformer models that power AI systems such as ChatGPT.

The platform also supports lossless audio streaming, a high-quality audio standard that is typically available through premium streaming services such as Spotify or Tidal.

On-device AI assistants and context awareness

Image caption: The new chip will allow devices to feature AI assistants capable of two-way conversations. (Image credit: Keumars Afifi-Sabet/Live Science)

Qualcomm also highlighted the platform’s “always-on intelligence.” An on-device eNPU AI accelerator can power virtual assistants, transcription, translation and context-aware features.

With context awareness, the system can continuously monitor a user’s environment and collect information to build a personal profile. Qualcomm says that profile could help the AI assistant answer future questions.

Although the system is focused on audio, it is designed to work with sensors and cameras, including earbuds with built-in lenses that can passively capture footage of a wearer’s daily life. That information could contribute to the data used by AI agents.

Privacy concerns surrounding camera-equipped earbuds

The prospect of wearable cameras raises privacy concerns. Qualcomm representatives said future earbuds could use low-resolution lenses and automatically apply AI-generated masks to obscure people, making it impossible to identify specific individuals.

The company has not yet disclosed how these features will affect battery life compared with conventional non-AI devices. Qualcomm representatives said manufacturers will not have to choose between sound quality, AI capabilities and battery life.

However, Andrew Leister, a staff product manager at Qualcomm, acknowledged during a media question-and-answer session that users may experience battery drain differently. Traditional earbuds are generally used intermittently for calls and music, while these new devices are intended to operate continuously.

How does Qualcomm’s chip compare with existing AI headphones?

AI-enabled headphones already exist to some extent. Viaim launched its Rise and RecDot devices this year, offering features for recording, transcribing, translating and summarizing conversations. The company has not publicly disclosed the chips that power those products.

Apple’s AirPods can also give users access to Siri, but complex tasks require a continuous connection to an iPhone.

Researchers have also explored smart audio systems. In 2024, scientists at the University of Washington worked on a system that could be integrated into off-the-shelf earphones to focus on a person speaking in a noisy environment while reducing background noise.

Another research project used deep-learning systems in headphones so wearers could use voice commands to control selective active noise cancellation and choose which sounds to hear or block.

Other companies are developing new audio technologies for headphones. Three years ago, xMEMS announced Cypress, an audio chip that generates ultrasonic waves through the piezoelectric effect. The chip became production-ready in September 2025, but no consumer devices based on the platform have been announced.

Apple may also be exploring similar technology. According to Bloomberg Editor-in-Chief Mark Gurman, the company could be working on AirPods with built-in cameras that collect data as users go about their daily lives.

Snapdragon Sound Elite Gen 2 performance improvements

The new chip is a successor to Qualcomm’s Snapdragon S7 Gen 1, which powered devices such as the Xiaomi Buds 5 Pro series. The earlier platform also used AI for active noise cancellation, audio personalization and lossless audio streaming while supporting low-power Wi-Fi connections.

According to Qualcomm, Snapdragon Sound Elite Gen 2 offers:

• Twice the AI processing power of the previous chip
• 128 billion operations per second, compared with 64 billion previously
• 40% lower power consumption
• A 30% smaller chip size

A key feature is the ability to store device apps in the chip’s onboard memory. Because the device has its own IP address, it can communicate with servers through the HTTPS protocol without routing the connection through a smartphone or computer.

Leister said Qualcomm is still examining whether the device can run two apps simultaneously in RAM—for example, an AI agent and an audio streaming service—while storing other apps in flash memory.

Qualcomm previously had to address the challenge of making low-power Wi-Fi consumption comparable to Bluetooth. The next challenge is running apps directly on the wearable device, Leister said.

“This is like the first smartphone moment when the Play Store came out, but in this sense, the app store for earphones is now owned by the device manufacturer,” Leister said. “They choose which apps to pre-bundle and we work from there.”

When will the new AI audio chip arrive?

Qualcomm has not disclosed when Snapdragon Sound Elite Gen 2 will appear in consumer devices. Company representatives said Qualcomm is considering working with companies including HP and Bose to develop headphones and speakers using the chipset.

Live Science was invited to Snapdragon Summit 2026 courtesy of Qualcomm, which covered travel and accommodations.