Yandex.Taxi to Implement Facial and Voice Recognition for Driver Authentication

Yandex.Taxi has announced plans to introduce facial and voice recognition algorithms in its driver app. According to the company’s blog, the program is already operating with 90% accuracy.

Currently, when launching the “Taximeter” app, drivers must enter their phone number and authenticate using a one-time code sent via SMS. However, Yandex.Taxi is preparing to offer a more convenient and secure authentication method, where a driver’s identity will be confirmed by unique characteristics—specifically, their facial features and voice.

“The main advantage of this kind of key is that it’s always with you: it can’t be faked, forgotten at home, or lost,” the developers wrote. To log in, a driver will need to take a photo using their smartphone or tablet camera and read a short phrase displayed on the screen. Yandex.Taxi promises that this process will take no more than a minute.

In developing the biometric authentication system, the company used its own image recognition technology, which is currently used in the Yandex search engine for image searches, as well as the SpeechKit voice recognition algorithm that powers the “Alice” voice assistant.

The program has already demonstrated over 90% recognition accuracy, and the company continues to test it with drivers. “Our goal is to train the model to recognize a person in any situation: even if they take a photo in poor lighting with a low-quality camera, have a hoarse voice, or, say, have grown a mustache,” the authors noted.