Explore Online Continuous Interactive English Speech Recognition Based on the Combination of Artificial Intelligence and Big Data

This paper explores the challenges faced by machine translation technologies in practical applications in recent years. Current machine translation technologies lack accuracy and significantly lag behind human translations, leading to an urgent market demand for a high-performance, high-accuracy machine translation technology. Consequently, this study introduces an online continuous interactive English speech recognition method based on the NOSE algorithm, constructing a continuous interactive recognition system. This system employs a knowledge database of expert information and a reference database for speech correction and evaluation, utilizing the NOSE algorithm to score and correct pronunciation errors. The recognition module is designed to consider the processing capabilities of hardware devices, using technologies such as DSP signal processing chips. Additionally, the design of the speech recognition function incorporates methods like the Mel Frequency Cepstral Coefficients and Fourier Transform to enhance the efficiency and accuracy of speech recognition. Comparative simulation tests and experimental results validate the performance of the proposed method in English translator speech recognition applications. The experiments demonstrate that this method surpasses existing methods in terms of speech recognition accuracy and timing, offering substantial application value.