HOTWORD DETECTION ON MULTIPLE DEVICES

Smart overview of the Invention

Methods and systems for detecting hotwords across multiple devices are proposed, enabling a seamless interaction in speech-enabled environments. The process begins when a first device receives audio data corresponding to a spoken utterance. It then calculates a likelihood score indicating whether the utterance includes a designated hotword, while a second device performs a similar calculation.

Comparative Analysis of Scores

The first device receives the likelihood score from the second device and compares both scores. Based on this comparison, if the first device has the higher score, it will proceed to initiate speech recognition processing on the audio data. This ensures that only the most relevant device responds to the user's command, enhancing user experience.

Importance of Hotword Recognition

In a connected environment where multiple devices are present, it is crucial to identify which device should respond to a user's command. The use of a predetermined hotword allows the system to filter out irrelevant utterances and focus on those directed specifically at it. This mechanism is vital for preventing multiple devices from responding simultaneously, which could lead to confusion.

Network Communication Between Devices

The system facilitates communication between devices over local networks or short-range radio, allowing them to share their computed likelihood scores. This collaborative approach enables devices to collectively determine which one should remain active for further processing of audio data, based on comparative confidence levels.

Advantages of the Proposed System

This innovative method provides several advantages, including improved accuracy in recognizing user commands and minimizing unnecessary activations across devices. Ultimately, it streamlines user interactions with technology in various settings, ensuring that only the intended device responds to voice commands while maintaining a smooth operational flow.