A Sound Scene Recognition Method Based on Label Augmentation and Multispectrum Fusion

Abstract

The invention discloses a sound scene recognition method based on label amplification and multi-spectrogram fusion, which includes: using different signal processing technologies to generate multiple spectrograms for sound scene data; The convolutional neural network model is used as the basic classification model; the label amplification technology is used to expand the super-category labels for the samples, and the artificially constructed hierarchical labels are used to improve the original network model into a multi-task learning model, and optimize the basic classification model. Performance; using the improved basic classification model to extract sample features, splicing multiple depth features of sound scene files, and reducing dimensionality to obtain global features. Fuse multiple global features corresponding to different spectrograms, train the SVM classifier, and use it as the final classification model. The invention applies the multi-spectrogram feature fusion technology to effectively improve the recognition performance; the proposed label amplification and model improvement method can effectively optimize the performance of the basic classifier and can be extended to other application researches.

Description

technical field [0001] The invention belongs to the technical field of scene recognition, and in particular relates to a sound scene recognition method based on tag amplification and multi-spectrum fusion. Background technique [0002] The sound scene recognition technology analyzes the audio data to determine the attributes, functions and uses of the space environment where the machine is located. Sound scene recognition based on convolutional neural network has become one of the most effective methods in this field. Since the sound scene dataset is labeled according to the function of the place, the problem of similarity between classes is more prominent, such as libraries and self-study classrooms, which are easy to misjudge each other. On the other hand, these data, which are inherently similar in acoustic features, are indiscriminately considered as different categories when training the network model due to different functional purposes, which hinders the network mode...

AI Technical Summary

Problems solved by technology

Similarly, Document 4 assumes that hierarchical labels already exist, and does not consider how to cluster and extract supercategory labels

Images