Corpus acquisition method based on neural network

In the wide application of computer technology and multimedia technology, corpus is widely used in linguistic theory research, military field and so on. Especially in the era of big data, the branch of language research with the corpus method as the core has been attached great importance by scholars all over the world, and a large number of corpora have emerged one after another. After scientific selection and effective annotation, corpora with appropriate scale can fully demonstrate and record the use of language, and people can grasp language facts by observing corpora. This paper analyzes the operation rules of language system and provides convenient conditions for language research. On the basis of understanding the current situation of corpus design and application in the new era, this paper mainly explores the speech emotion feature extraction method based on deep neural network. The final experimental results show that the classification of emotion features by extracting statistical features has a good recognition effect, and the corpus acquisition method based on neural network has a large development space.