Abstract: The performance of spinning spindles decisively influences the production efficiency and yarn quality. This article analyzes the noise during the running process of spinning spindles. Sound pressure signals which collected during the experiment is analyzed and separated by spectrum analysis method and FastICA algorithm. Thus the source of the noise component during the running process of spinning spindles is separated and contrast with the main frequency of the motor noise. 38.88% of the main frequency is identical, 44.44% of the main frequency is close with the minimum error 0.05%. Experimental and analytical results show the effectiveness and correctness of FastICA algorithm used in sound pressure signals separation.