Voice wake-up method and device, electronic equipment and storage medium

Abstract

The invention relates to a voice wake-up method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a wake-up voice; the forward end-to-end wake-up system detects whether a wake-up event exists in the wake-up voice; if the forward end-to-end wake-up system detects that the wake-up event exists in the wake-up voice, the reverse end-to-end wake-up system detects whether a wake-up element exists in a reverse feature of the wake-up voice; and if the reverse end-to-end wake-up system detects that a wake-up event exists in the reverse feature of the wake-up voice, waking up corresponding equipment. According to the method and the device, the positive end-to-end wake-up system is awakened through the wake-up voice, and the reverse end-to-end wake-up system is awakened in combination with the reverse characteristics of the wake-up voice, so that the accuracy of the wake-up device is improved, false wake-up is reduced, and the user experience is improved.

Description

technical field [0001] The present application relates to the technical field of voice wake-up, and in particular, to a voice wake-up method, apparatus, electronic device and storage medium. Background technique [0002] In the voice interaction system, wake-up is an important interaction entry. Wake-up requires real-time voice processing, not only high wake-up rate and low false wake-up rate, but also demanding system resources such as memory and power consumption. The current wake-up system usually includes modules such as feature extraction, acoustic calculation, decoding, and confidence judgment. Usually, it is necessary to stream the voice input, extract the voice features frame by frame, send it to the input acoustic model to calculate the acoustic score, then transmit the acoustic score to the wake-up network for recognition, and finally output the wake-up confidence score through the confidence calculation. [0003] The existing voice wake-up system processes the o...

AI Technical Summary

Problems solved by technology

[0003] The existing voice wake-up system processes the original voice when performing acoustic calculation and decoding. Even if multiple systems are used for judgment, since multiple systems are trained from the same time-sequence voice training, the difference between the models There is a large similarity between them, and it cannot provide additional new information for false wakeup suppression; therefore, such systems usually have high false wakeups, which is also a difficulty in the current wake-up system

Images