Infant cry detection method, device and readable medium in real scene

Abstract

The present invention relates to a method and device for detecting baby crying in a real scene, and a technical solution of a readable medium, comprising: collecting audio data including baby crying in a real scene, labeling and pre-processing the audio data as a data set , to obtain the network input data; input the network input data to the deep neural network including the feature extraction network, the human voice detection network and the cry detection network, and carry out the training of feature extraction, human voice detection and cry detection respectively, and obtain the human The first loss function and the second loss function corresponding to the sound detection network and the cry detection network; the third loss function is obtained by training the deep neural network as a whole, and the baby cry detection model is obtained; The audio data collected in the scene is detected, and the baby crying detection result of the real scene is obtained. The beneficial effect of the present invention is that the baby's cry can be detected more accurately in a relatively short time.

Description

technical field [0001] The invention relates to the field of computer artificial intelligence, in particular to a baby crying detection method, device and readable medium in a real scene. Background technique [0002] For newborn babies, crying is their main way of communicating with the outside world. Babies express some physiological needs (such as hunger, thirst, physical discomfort, etc.) through crying. Now more and more smart home products provide the function of baby crying detection, which can assist parents to take care of their babies. [0003] There are still many deficiencies in the existing baby crying detection system. For example, the continuous crying generally takes more than 30s to be successfully detected, and the requirements for the application environment are relatively high. Most products can only achieve ideal results in a quiet room. The detection effect, when the ambient sound is slightly complex, the detection performance drops sharply. Content...

AI Technical Summary

Problems solved by technology

[0003] There are still many deficiencies in the existing baby crying detection system. For example, the continuous crying generally takes more than 30s to be successfully detected, and the requirements for the application environment are relatively high. Most products can only achieve ideal results in a quiet room. Detection effect, when the ambient sound is slightly complicated, the detection performance drops sharply

Images