Google announces edge face and voice biometrics at I/O Conference

Google has launched a new smart home hub with biometric facial recognition processed at the edge, and reduced its voice recognition models to support edge voice biometrics, among a series of announcements made by Google at its I/O Conference, and reported by TechCrunch.

Nest Hub is a rebranded Google Home Hub, with a price reduced by $20 to $129, while the Nest Hub Max has a 10-inch camera instead of the regular 7-inch version, as well as the camera enabling its “Face Match” feature. Face Match works with on-device machine learning, the company says. The Nest Hub Max MSRP is $229, and comes with a switch to disable the camera and microphone.

Google Cloud announced the suspension of its facial recognition offering in December, and committed not to resume selling it until it can avoid abuse of the technology and harmful outcomes.

Google Home and Nest product VP and general manager Rishi Chandra said the company is committed to not using data collected by Nest to personalize advertisements, the LA Times reports. The company is attempting to regain consumer trust lost when its Nest security hub was discovered to include a secret microphone earlier this year, according to the Times. Chandra called it a “mistake” to have not informed customers of the microphone when the device went on sale.

Smart home products and smartphones can also accommodate voice recognition, as Google has reduced its models from hundreds of gigabytes to half a gig. Edge processing can make conversations with Google Assistant nearly instantaneous, and enable them in offline and airplane mode. The next-generation Assistant will reach new Pixel phones later this year. Google announced the development of a new method of training models with an RNN-T to deploy on-device speech recognition earlier this year.

The company is also researching how to tailor its voice algorithms to understand users with speech impairments, in a new initiative called Project Euphonia.

Other announcements include new Pixel smartphone models, 3D augmented reality models for certain search results, new Google Lens capabilities, and web support for artificial intelligence-powered customer service application Duplex. Google Assistant and Maps have added several functions as well.

News for developers includes the promotion of Kotlin by Google as the primary programming language for Android developers, and support for games on the company’s smart display devices.

Article Topics

biometrics  |  biometrics at the edge  |  facial recognition  |  Google  |  on device matching  |  voice recognition