Nearby wake-up method and device for terminal, computing equipment and storage medium

Abstract

The invention discloses a nearby wake-up method and device of a terminal, computing equipment and a storage medium. The method comprises the following steps that: when a target terminal receives the wake-up word of a user, a target audio with a set duration is obtained; the target terminal calculates a wake-up voice signal-to-noise ratio according to the target audio and sends the wake-up voice signal-to-noise ratio to a central control device so as to indicate the central control device to determine a nearby terminal of the user according to wake-up voice signal-to-noise ratios sent by at least one terminal and send a wake-up permission instruction to the nearby terminal; and if the target terminal responds to the user if receiving the wake-up permission instruction. According to the method provided by the embodiments of the invention, the wake-up voice signal-to-noise ratio is used as the selection basis of the nearby terminal, and calculation complexity is relatively low. The methodhas higher accuracy compared with a method judging the energy value of a wake-up audio; the wake-up voice signal-to-noise ratio is only needed to feed back to the central control device; requirementson the stability of a network are relatively low; the real-time performance of the method is relatively good; and user experience is improved.

Description

technical field [0001] Embodiments of the present invention relate to data collection and processing technologies, and in particular, to a nearby wake-up method, device, computing device, and storage medium for a terminal. Background technique [0002] At present, after the user enters the wake-up audio, the method of screening the nearest smart devices among multiple smart devices is relatively common. The existing smart device wake-up methods are mainly divided into two types. The first is to calculate the energy information of the wake-up audio. And introduce the device compensation factor and environmental impact factor to correct the energy value of the wake-up audio, and send the energy value of the wake-up audio corrected by each smart device to the third-party device, and the third-party device will filter the nearest smart device according to each energy value; One is to calculate the confidence level of the wake-up audio, the distance and direction between the smar...

AI Technical Summary

Problems solved by technology

[0003] Although the above methods can select the nearest smart device among multiple smart devices, the first method cannot adapt to new smart devices and application scenarios after using the preset device compensation factor and environmental impact factor, which increases the complexity of the algorithm Secondly, when the energy value is used as the judgment basis, the influence of the device on the wake-up audio energy information when playing its own audio is not considered, which may easily cause misjudgment by the third-party device in the actual use environment; the second method needs to calculate and transmit multiple Parameters have high requirements on the memory, computing speed and network stability of smart devices; secondly, using neural networks to perform feature fusion analysis on multiple parameters, the real-time performance is poor, which reduces the user experience

Images