# The discriminating power of experienced listeners

Consider a choir of n singers who sing the same tune in almost perfect unison.

When asked how many singers they do hear, test persons may say:

• don't know

• more than 3, 5, 10, ...

• 3, 4, 5, 6, 7, ...

At the heart of discriminating and thus counting singers just by listening lies the ability to distinguish voices by timbres, i.e. to analyse a highly complex frequency spectrum. Experienced listeners will typically perform better than unexperienced ones. Disturbing effects of reverb and so on must be taken into consideration (and possibly be minimized or even eliminated).

Are there reported "world records" like

X was able to reliably tell if there were 3, 4, 5 up to 12 singers singing?

Is there scientific literature on this topic?

You may want to compare these questions to this seemingly more natural one:

Let n voices play n different tunes (as in a big band or any kind of orchestra) - how many voices can be distinguished (by the most experienced listeners)?

• I'm unaware of any such literature, and it's hard to imagine any way of getting valid results, because there are so many variables that are not controlled. No group of singers sings in perfect unison, and every group varies from perfect unison in different ways. Unless you set up a contest where there is only one recording to judge from, you will get different results every time. Jan 22, 2020 at 15:45
• I think when 2 or more voices sing in perfect unison ... the result should be one single voice with an new timbre: the sum of all formantes. Jan 22, 2020 at 16:45
• @AlbrechtHügli: Is this so? In this case, the unison should be only almost perfect. (I slightly changed the question.) Jan 22, 2020 at 16:47
• This is just an assumption! But I think: If John and Paul and George sang unisono I hear a mix of them and supposed I know them each one I could probably identify the one or the other or I can say when I miss one of them. Before I‘d ask how many voices there are I would ask which variables can identified. Jan 22, 2020 at 17:47
• "At the heart of discriminating and thus counting singers just by listening lies the ability to distinguish voices by timbres" -- where is this assumption coming from? // The ability to estimate the number of performers depends on many additional variables. How are you going to control for all of them? Jan 22, 2020 at 18:23