Apple has largely avoided discussing modern AI systems, but the company’s voice recognition service, Siri, is exploring new capabilities. A recent patent application describes a system that uses motion data to improve voice commands. The system is designed to analyze users’ mouth movements to enhance voice recognition, without relying on a camera or additional sensors that drain battery and processing power.

Instead, it utilizes the device’s motion sensors, such as accelerometers or gyroscopes, to detect mouth, neck, or head movements that could indicate human speech. To support this system, Apple would need extensive data on how humans use their mouths. The company could establish voice profiles for users and utilize the Live Speech feature on iOS to record voice profiles for speech-to-text processing.

In addition, Apple discusses the training of a “first language model” for facial movement recognition, hinting at the potential use of machine learning. While not all of Apple’s ideas come to fruition, this recent development seems to carry more weight than some of the company’s previous concepts. It aligns with Apple’s strategy of integrating AI into new features, suggesting a potential evolution for Siri and voice recognition technology in Apple products.