One microphone recording of train pass-by noise contains the contributions of all noise sources, i.e. traction noise, rolling noise, aerodynamic noise, and less important sources such as train body vibration noise. As rolling noise can now be predicted accurately and traction noise is negligible at high speed, one microphone recordings of high-speed train noise become useful for empirically modeling railway aerodynamic noise. In this work, rolling noise of Swedish X2 high-speed trains was described using the total transfer function and the total roughness, which were determined using the indirect roughness method based on measurements made in 2004 by the SP Technical Research Institute of Sweden (SP). One microphone data of X2 train pass-by noise were collected by SP in 1994; these data cover a wide speed range from 70 to 270 km/h and are therefore useful both for empirically modeling aerodynamic noise and for validating the modeling of rolling noise. This work also confirms that railway aerodynamic noise generally has dipole characteristics. However, it was discovered that at 250 Hz and below monopole sources dominate railway aerodynamic noise.