Abstract: This paper investigates energy distribution of air-borne acoustic signals in time-frequency domain during aluminum alloys spot welding in order to identify weld nugget states. Experiment results indicate that air-borne acoustic signals contain information about weld nugget states and can be viewed as a kind of parameters that indicate the welding quality. Methods of time-frequency analysis are put forward to analyze the non-stationary signals. The results of experiment and calculation show that Choi-Williams distribution, a kind of time-frequency analysis, has been proved to be a better approach to discriminate weld nugget states by analyzing the signals. As a result, it can be employed to recognize signals in the industrial application and can provide a new way for various welding processes.