Hybrid algorithm for English speech recognition based on cloud computing technology

With the frequent growth of international exchanges, the communication between languages is very important, especially as English is one of the main international communication languages, English speech recognition is an important research field in natural language processing. With the rapid development of cloud computing technology in recent years, English speech recognition algorithms have begun to use cloud computing technology to process speech recognition tasks. This paper proposes a hybrid algorithm for English speech recognition through cloud computing technology combined with English speech program. English speech contains rich language and literature. In many international conferences, the accuracy of English speech recognition directly affects the presentation of conference content. Therefore, English speech recognition is very important for communication between different language users. This paper studies the modeling method of end-to-end speech recognition in the case of limited English speech recognition data. We mainly focus on two hybrid algorithms for end-to-end speech recognition: connection temporal distribution (CTC) and attention-based encoder-decoder network (LAS). Research to improve the encoder performance of the connection timing distribution model and improve the performance of the encoder based on the attention mechanism, the final result of the research shows that when the speech rate is 90 words per minute, the accuracy of speech recognition is 94.52%. It can be seen that when the speed of speech increases, the accuracy of speech recognition will drop slightly, but the overall rate remains above 94%, which is highly feasible.