Hot word-free preemption for automated assistant response presentation

• Main Authors: VAN DER RIANTO ALEX, CHAO PU-SEN
• Format: Patent
• Language: Chinese; English
• Published: 30.12.2022
• Subjects: ACOUSTICS; MUSICAL INSTRUMENTS; PHYSICS; SPEECH ANALYSIS OR SYNTHESIS; SPEECH OR AUDIO CODING OR DECODING; SPEECH OR VOICE PROCESSING; SPEECH RECOGNITION

Responsive to a hotness-free utterance received during presentation and determined to be likely directed to the automated assistant, presentation of an automated assistant response may be selectively preempted. A determination may be performed that an utterance may point to an automated assistant, for example, using an utterance classification operation performed on audio data received during presentation of a response, and based on such a determination, a response may be preempted with another response associated with a later received utterance. In addition, a duration for determining when a conversation between the user and the automated assistant should be terminated at the end of the conversation may be dynamically controlled based on when the presentation of the response is completed. 响应于在呈现期间接收到的并且被确定为可能指向自动化助理的无热词话语，可以选择性地抢占自动化助理响应的呈现。可以例如使用对在响应的呈现期间接收的音频数据执行的话语分类操作来执行话语可能指向自动化助理的确定，并且基于这样的确定，可以用与稍后接收的话语相关联的另一响应来抢占响应。另外，可以基于响应的呈现何时完成来动态地控制用于确定在用户与自动化助理之间的对话结束时应该何时终止会话的持续时间。