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

Abstract

The embodiment of the invention provides a voice wake-up method and device, electronic equipment and a computer readable storage medium, and the method comprises the steps: carrying out supervised training of a voice sample marked with a text tag and a classification tag for judging whether the voice is wake-up word voice or not, and obtaining a multi-task voice wake-up model, therefore, the multi-task voice wake-up model in the embodiment not only can output the predicted phoneme sequence of the voice, but also can output the probability of whether the voice is the wake-up word or not (i.e., the wake-up probability). According to the method, whether awakening is carried out or not is jointly judged by utilizing the two factors of predicting the phoneme sequence and the awakening probability, and compared with a traditional single-factor (phoneme sequence) judgment method, the awakening probability can be considered at the same time even if an accurate phoneme sequence cannot be obtained in a noise environment, so that whether awakening is carried out or not is determined. Therefore, the awakening accuracy in the noise environment can be improved.

Description

technical field [0001] Embodiments of the present invention relate to the field of voice processing, and in particular, to a voice wake-up method, device, electronic equipment, and computer-readable storage medium. Background technique [0002] Voice wake-up technology is an important branch of voice recognition technology. Currently, it has important applications in vehicle equipment, voice navigation, voice assistants, smart homes, etc., and is used to start programs or services through voice. In the voice wake-up technology, a voice wake-up model is usually used to predict whether to wake up a corresponding program or service. The traditional voice wake-up method uses a single-task phoneme sequence prediction model to obtain the phoneme sequence of the voice according to the acoustic characteristics of the voice signal, and wakes up when the predicted phoneme sequence matches the phoneme sequence corresponding to the wake-up word. The accuracy rate is higher in a clean e...

AI Technical Summary

Problems solved by technology

The traditional voice wake-up method uses a single-task phoneme sequence prediction model to obtain the phoneme sequence of the voice according to the acoustic characteristics of the voice signal, and wakes up when the predicted phoneme sequence matches the phoneme sequence corresponding to the wake-up word. The accuracy rate is high in a clean environment, but when the background noise is large, the wake-up accuracy rate is low due to the inability to accurately obtain the phoneme sequence of the voice

Images