SYN-P: A Simulated Dataset of Friedlander Blast Sound Propagation

The SYN-P dataset provides simulated blast sound samples modeled after the Friedlander waveform (0.5s duration, 48kHz sampling rate). The data is synthesized based on a physical parameter grid comprising the characteristic period T_ID (ranging from 0.8 ms to 4.0 ms) and the propagation distance r (ranging from 20 m to 25 km). The generation process begins with amplitude spectrum modeling that accounts for energy attenuation, followed by an IFFT with a random phase array to transform the signal into the time domain. Subsequently, the signal is shaped using an attack/decay envelope and a dynamic (time-varying) low-pass filter to faithfully recreate the dissipation characteristics of real-world acoustic propagation. Finally, all data is amplitude-normalized, window-tapered to prevent click noise, and packaged into .mat files.