System and method for selective active word detection using neural network model

摘要

Systems and methods for playing media through a media playback system include capturing sound data through a network microphone device and identifying candidate active words in the sound data. Based on the identification of candidate active words in the sound data, the system selects a first active word engine from a plurality of active word engines. Through the first active word engine, the system analyzes sound data to detect the identified active word, and in response to the first active word engine detecting the identified active word, one or more remote computing devices associated with the voice assist service. The voice utterance of the sound data is transmitted.