Design of Intelligent Translation English Corpus Based on Deep Learning

In order to further improve the effect of English intelligent translation, this paper combines the idea of deep learning to build an intelligent corpus system. Aiming at the problem of a lot of noise in speech data, the speech data is preprocessed by Mel frequency cepstrum coefficient method, and then transformers are combined with mutual information to recognize speech emotions. Then, by integrating diverse data sources, a wide range of English text data is provided to support complex translation and processing tasks. During the data acquisition phase, English text is collected from online platforms and social media through automated tools. In the data processing stage, advanced text processing technology is used for text analysis and preprocessing to ensure the quality and consistency of data. Finally, this paper verifies the effectiveness of the model proposed in this paper with experimental analysis. In the future, intelligent English translation subjects should take a positive attitude towards science and technology, and regard intelligent translation technology as an effective tool for English translation.