High-speed information processing method based on big data

Abstract

The invention provides a high-speed information processing method based on big data. The method comprises the steps that for each frame audio of an audio block, the parameters of the last layer in a convolutional neural network are updated, namely a weighting coefficient W and offset b; a BP algorithm is used to update the parameters of other layers in the convolutional neural network; the polynomial representation of each frame audio is updated according to new parameters; and based on the existing voice library D, the current audio block is reconstructed and an error epsilon is calculated, namely, the polynomial representation Yk of the current audio block Xk is reconstructed. The method provided by the invention is based on voice big data processing, and has the advantages of stronger anti-noise performance, higher accuracy rate and higher recall rate. The knowledge acquisition efficiency of a user is significantly improved.

Description

technical field [0001] The invention relates to voice data processing, in particular to a high-speed information processing method based on big data. Background technique [0002] With the advancement of science and technology and the development of network technology, there is a large amount of available media information in the network, such as call records, WeChat voice, and meeting records. Faced with a large amount of audio data, users need to understand voice information faster, save user time, and improve work efficiency. With the rapid development of information retrieval technology, speech summary generation technology is becoming more and more mature. From the initial method based on word frequency to the introduction of machine learning, the performance has been greatly improved. Existing schemes generally use supervised learning methods. In the training set, the classification model is used to train the training set to obtain the optimal weighting coefficient v...

AI Technical Summary

Problems solved by technology

Existing schemes generally use supervised learning methods. In the training set, the classification model is used to train the training set to obtain the optimal weighting coefficient vector, and to classify and predict the test set; however, relying on the supervised learning model requires labeling data, usually by Manual labeling is very time-consuming and subjective, and it is easy to ignore the semantic similarity between sentences, which reduces the accuracy of calculation results

Images