Yes, it’s the 2010 World Cup. Played in our brand-new stadiums.
Now, since I find that the ‘beautiful game’ is about as interesting as watching my dog chewing the carpet, I won’t be watching. Hundreds of thousands will. Now, let’s imagine that there is an emergency in the middle of one of the matches. The crowd panics.
The organisers try to keep control, but it’s no good. In the panic, people are injured and some die.
Now, let’s say that some person who has been injured wants to sue the stadium’s owners. One sure way of winning is to claim that the stadium’s public-address system was not intelligible enough for the crowd to hear instructions and so the crowd went out of control – and disaster struck.
A way of making sure that this does not happen is for each public-address system to be tested to a standard so that it can be claimed that the system transmits intelligible speech. The subject of speech intelligibility is not simple. I have been in conference centres where the opera- tors have assured me that “everybody can hear fine”. Thus, we have the operator, who is in no sense an expert, giving expert judgment. This will stand up in a court of law in the same way a mirror stands up to a blow from a trip hammer.
The first speech intelligibility paper which I read was written by Victor Peutz in 1958. He proposed that speech intelligibility is a function of the degree to which we can hear consonants. Take this sentence: “The birch canoe slides down the smooth plank.” Now, read it without the vowels: “Th brch cn slds dwn th smth plnk”. While this may sound like your girlfriend’s SMS on mixxit, it still reads better than the sentence without consonants: “e i aoe ie o e oo a” – which sounds like my dog attacking the cat.
But you get the picture – speech intelligibility can be related to something tangible and, in the case of Peutz, it is when we don’t hear the consonants in words, largely owing to the reverberation time of an area.
There is another way of looking at the matter. This is based on RaSTI, STI or STI-PA. The acronym RaSTI stands for ‘rapid speech intelligibility index’. We can break the male or female voice into fundamental frequencies and then we assume that each of these frequencies is modulated, by actual speech. The modulation is in the range from 0,2 Hz to 12,5 Hz. So what one could do would be to program an audio generator to generate frequencies over bands from 120 Hz to 1200 Hz and modulate each of the bands from 0,2 Hz to 12,5 Hz in succession. One sets up the generator with a loudspeaker in a venue (hall, stadium). One could then have a meter having filters which measure the modulated signal all over the venue. If the modulated signal measured all over the venue is the same as that transmitted from the speaker, then you could say that the speech intelligibility is good. This measurement, that of the STI (or the speech transmission index) takes about 98 measurements to complete and is impractical for complex spaces as, if you just measured in ten places, you would have to take 980 measurements.
A simple method is STI-PA, which stands for ‘speech transmission index for public-address systems. This is based on the STI, but is simpler.
The measurement of speech intelligibility is covered in the UK and Europe by IEC and BS standards and in the US by the NFPA. As far as I can establish, there is no local standard other than the earnest assurances of the suppliers of the PA system that “sure, it works”.
There are firms in South Africa that can measure speech intelligibility, but I never see their results being referred to. With the sole exception of the Newlands cricket stadium, at all the spaces I go to (conference centres, stadiums and airports), the public-address systems are not very intelligible. One wonders why public-liability insurers don’t take this up. But, perhaps, they don’t like spectators of “the beautiful game” either.