Perhaps tomorrow I might experiment with snapping some stone tiles in two and seeing what that looks like.
Here a stone tile was suspended between two bricks. A 170lb dynamic weight (me) was applied quickly but gradually. The tile failed with a sharp crack. Analysis shows this was composed of multiple events due to the proximity to the ground.
2 frames (about 0.01 seconds) later, first crack noise has decayed:
Second crack, plus other noise.
Ground impact of wood & tile
Sound has decayed:
So the initial snap of the tile was over in less than 0.01 seconds, The entire event took less than 0.04 seconds.
It seems entirely reasonable that a rapid serious of large scale crackings would be one possible explanation for the faint rapid series of cracks heard at the onset of the Plasco collapse.
[Minor tech note for completeness, the video was recorded at 240 fps, so each frame is spread somewhere within 0.0042 seconds, with different portions of the frame at different times due to the rolling shutter. So it's impossible to pinpoint an exact time for each frame. However this means, at most, an error of half a frame, or 0.002 second. The waveform and the image sequence still tell the story very well, with the important thing being the initial crack which would be isolated in a larger structure]