How loud can a woman scream?


The most critical variable to consider in doing an acoustical analysis of the screams that were heard in the early hours of  14 February 2013 is how loud Reeva could have screamed. Since no one ever measured Reeva screaming under the circumstances that she faced that night we are left to make assumptions about how loud she could have screamed.

The defence’s acoustic expert, Ivan Lin, based all his analyses assuming that Reeva and Oscar would have screamed at a sound level of 110 dBA.

“From my experience and the past researches, typical human screams recorded as between 110 dBA [100dBA] to 120 dBA.  For the purpose of this exercise, I shall use an average value of 110 dBA.” (Page 2643, Lines 6–9) [the 110 dBA is an incorrect recording – Lin actually said 100 dBA]

Read here for more information about sound levels and how they measured.

His assumption that Oscar would have screamed at the same level as Reeva is already problematic given the wealth of evidence that women on average scream substantially louder than men. In addition, a woman locked up in a toilet fearing for her life is unlikely to produce a typical average scream.

Making an assumption about how loud Reeva would have screamed cannot be made lightly or arbitrarily. It may look small on paper, but 10 dBA is a huge difference on the logarithmic decibel scale. A scream at 120 dBA sounds more than twice as loud to the human ear as a scream at 110 dBA.

Let’s do a very simple and easy exercise. Please open Google Search and type in the phrase “how loud can a woman scream”? Within the top three results you will find the following link:

http://www.audioforensics.com/PDFs/AES122_Scream.pdf. (Forensic Analysis of the Audibility of Female Screams by Dr Durand Begault, AES 33rd International Conference, Denver, CO, USA, 2008 June 5–7)

In a controlled soundproof environment Dr Begualt asked ten females to scream as loud as they could, as if they have been surprised by something scary. The maximum sound levels achieved in rank order are: 123, 122, 122, 118, 115, 110, 109, 109, 108 and 102 dBA’s. The three highest screams were all woman below 30 years old. And the sound level decreased with age. The 102 dBA scream was from the oldest participant, and at the time of the test she suffered from an allergy.

Reeva Steenkamp was 29 years old when she died.

What did the acoustic expert say in court? During cross-examination Advocate Gerrie Nel wanted to know if a woman fearing for her life could have put the dB closer to 120 than 110 dB. The expert said that 120 dB is almost the same as a jet engine taking off and that there is only a very slight possibility that she could have screamed at 120 dBA.

[Nel] That is so. If somebody screams in absolute fear, would you not put it closer to 120 than 110?

[Lin] I think if one need to appreciate what 120 means. 120 dB is extremely loud, it is almost hearing a jet taking off at a 100 meter away and that is not ... that is extremely [indistinct] human voice and it is a very slight possibility but it is a possibility.

[Nel] I appreciate what you just said. It is at the extreme limits of a person’s voice.

[Lin] Yes.

Page 4668: Lines 14 to 40.

Compare this “very slight possibility” and “extreme limits” with Begualt’s sample where 30 % of test subjects screamed louder than 120 dBA – and all of them were less than 30 years old.

It is common practice to, when dealing with variables and uncertainties, to perform a sensitivity analysis. This is what the expert did with the ambient (background) noise levels when he assumed a range of 20 dBA to 45 dBA. This while he could have gone out to the Silver Woods Country Estate at night and actually measure the real ambient sound level himself. The question is why didn’t he do a sensitivity analysis with the loudness of a scream? He said that a typical scream is between 100 dBA and 120 dBA. It would have been far more beneficial to the court if he performed a sensitivity analysis with both these different scream sound levels.

What is clear and what our own analysis will show is that any analysis performed with 120 dBA scenario would have been very detrimental to the defence’s case.

What did the court thought of the expert’s analysis and testimony?

In any event, the evidence of Mr Lin, an acoustic engineer, cast serious doubt on whether witnesses who were 80 metres and 177 metres away respectively from the accused’s house would be able to differentiate between a man and a woman’s screams, if the screams were from the toilet with closed windows.

(The Stipp’s were in fact 72 meters away from Oscar’s bathroom window – and not 80 meters as assumed by the expert and accepted by the court.)

We will show that the court’s conclusion above has no scientific merit and that the four state witnesses (Dr Stipp, Mrs Stipp, Mr Johnson and Ms Burger) would have heard Reeva’s scream audibly and intelligibly to the extent that they would have been able to tell the difference between a male and female scream. We will also show that the four state witnesses would have heard Reeva scream louder and clearer (i.e with greater quality, audibility and intelligibility) than the immediate neighbours and if there was a case of mistaken identification then the mistake was much rather made by the defence witnesses (the immediate neigbours) than by the State witnesses.

In closing lets remind ourselves of what the duty of an expert witness is:

“... the value of an expert is not to espouse and further the cause of a particular party, but to assist the Court in coming to a proper decision on technical and scientific matters. It should therefore at all times be remembered that an expert is primarily there to assist the Court and not necessarily to further the cause of his particular client to such an extent that he loses objectivity and in fact undermines his client’s case.”

State vs Huma

Whom did the acoustic expert serve?