This paper focuses on detecting real-life events reported on Twitter by analyzing the text stream. Twitter, as a social media platform, is rapidly emerging as a source of information for various events, but the challenge lies in distinguishing meaningful events from the vast majority of trivial tweets. The paper introduces EDCoW (Event Detection with Clustering of Wavelet-based Signals), a method that uses wavelet analysis to build signals for individual words, filters out trivial words, and clusters the remaining words to detect events. The method is designed to be scalable and effective, as it does not require pre-setting the number of events and can handle the dynamic nature of Twitter data. Experimental results show promising performance, with EDCoW achieving high precision in detecting relevant events. The paper also discusses the limitations of existing methods and highlights the advantages of EDCoW, such as its ability to automatically determine the number of events and its robustness to trivial tweets. Future work includes improving the method by incorporating semantic relationships between words, expanding the dataset size, and exploring user relationship analysis.This paper focuses on detecting real-life events reported on Twitter by analyzing the text stream. Twitter, as a social media platform, is rapidly emerging as a source of information for various events, but the challenge lies in distinguishing meaningful events from the vast majority of trivial tweets. The paper introduces EDCoW (Event Detection with Clustering of Wavelet-based Signals), a method that uses wavelet analysis to build signals for individual words, filters out trivial words, and clusters the remaining words to detect events. The method is designed to be scalable and effective, as it does not require pre-setting the number of events and can handle the dynamic nature of Twitter data. Experimental results show promising performance, with EDCoW achieving high precision in detecting relevant events. The paper also discusses the limitations of existing methods and highlights the advantages of EDCoW, such as its ability to automatically determine the number of events and its robustness to trivial tweets. Future work includes improving the method by incorporating semantic relationships between words, expanding the dataset size, and exploring user relationship analysis.