Regular expressions and finite state automata have been widely used in programs for pattern searching and string matching. Unfortunately, despite the popularity, regular expressions are difficult to understand and verify even for experienced programmers. Conventional testing techniques remain a challenge as large regular expressions are constantly used for security purposes such as input validation and network intrusion detection. In this paper, we present a lightweight verification framework for regular expressions. In this framework, instead of a large number of test cases, it takes in requirements in natural language descriptions to automatically synthesize formal specifications. By checking the equivalence between the synthesized specifications and target regular expressions, errors will be detected and counterexamples will be reported. We have built a web application prototype and demonstrated its usability with two case studies.