Reflections on the Pistorius trial

28th March 2014 By: Terry Mackenzie-hoy

I had hoped (really) to not write about the Oscar Pistorius trial but so many people have asked me (as an acoustics engineer) about noise propagation (of screams, gunshots, cricket bat noise) that I will settle it all, once.

Witnesses state that they heard screams and gunshots from Pistorius’s house in the early hours of the morning. Screams were of a woman’s voice. Is this possible and could the screams have been from a man and the gunshots mistaken for the sound of a cricket bat striking a door frame?

Firstly, could the screams be heard? The distance involved is 177 m, which is about the length of one-and-three-quarter rugby fields. In a ‘screaming competition’ held by a local school, I measured the screams of some 40 girls. The loudest was 104 dBA (decibels A-weighted) and the average at about 95 dBA. For a scream to be clearly audible, it has to be about 6 dBA louder than the ambient (background) noise level. A simple calculation, assuming a straight line of noise propagation, gives that, in the absence of wind, temperature inversion and rain, 95 dBA will be 50 dBA 177 m distant.

Thus, if the ambient noise level is 44 dBA, the scream will be clearly audible. The value of 44 dBA would very much be the correct ambient level for Silverwoods Estate. Further, we (my engineers) computer-plotted the noise contour (using the German program SoundPLAN 7.2) on a geomap of Silverwoods Estate, which confirms the above.

If the screams came from a person inside a toilet with the door closed, the noise would be about 28 dBA and the screams would not be audible. Since the witnesses say they heard the screams, it implies they did not come from inside a toilet with the door closed. Additionally, the screams could have come from another part of the area but, given the circumstances, very unlikely.

Secondly, could the screams be of a man? The answer is that it depends on the man. If you listen to cricketers, when a player is caught LBW, there is some feminine in the screams. But screams of a woman are pitched much higher than a man’s, which identifies them.

Thirdly, could the gunshots have been mistaken for the sound of a cricket bat striking a door frame? Oddly, if there was just one bat strike or one gunshot, yes. But multiple gunshots, no. A bat striking wood will make a noise of about 114 dBA. A gunshot (according to the US Federal Bureau of Investigation) is 145 dBA. Owing to the logarithmic nature of the decibel scale, a difference of 30 dB means a gunshot is much louder than a cricket bat. When you hear the first sound, you are alerted but do not know what you have just heard. The subsequent shots allow identification.

There is another issue: the gunshots will result in multiple echoes off buildings until the noise dies down like thunder. Cricket bat? No way.

Oh, yes, I know you have seen the YouTube video where a man simulates the incident by first chopping on the door frame with a cricket bat and then shoots through the door. This is translated as “man uses home video to record sounds and then you hear them played back through TV speakers”. This is similar to trying to hear if the Hallelujah chorus is the same volume as Brahms lullaby on your smartphone.

There are a few things which are peripheral to the trial. We can see that the courtroom has quite acceptable acoustic cladding in the form of variable-width slat absorbers on the walls and an acoustic tile ceiling. This is the Pretoria High Court, after all. A trip to any of the many magistrates’ courts around South Africa will show that the acoustics of these venues are bad, to say the least. Forget about justice being seen to be done – it should be heard to be done. But this is not the case. Many of our common criminals must still be wondering what the magistrate said. This is more shameful than the standards of translation and translators of the High Court and that is really saying something.