Posted on 17 May 2018
"What the Yanny/Laurel clip demonstrates is that speech is a complex set of sound frequencies that occur simultaneously, and which are combined in different ways to produce the great variety of speech sounds that exist in the world's spoken languages.
"As we age the sensitivity of our hearing to different parts of the frequency spectrum changes, particularly in the higher frequency bands. Just as with our vision, our hearing is susceptible to all sorts of intriguing illusions.
Age and ethnicity
"We hear speech differently depending on what we're looking at, and it's been shown in more recent experiments that the way we hear certain speech sounds is influenced by how old we judge the speaker to be, his/her perceived ethnicity, and even the clothes the speaker is wearing.
"The Yanny/Laurel clip is a very neat analogy to the 'blue/black or white/gold dress?' illusion, but in the domain of hearing rather than vision.
"In the phonetics teaching we offer in the Department at York we explore how the human ear detects sound, how those auditory signals are processed, interpreted and stored by the brain, and how they may be influenced and interfered with by information coming into the brain via our other organs of perception.
"We're also interested in how the perception of speech differs from the perception of other sorts of sound: speech, it seems, is handled differently from non-speech sounds at quite a basic cognitive level. Part of our job as forensic speech scientists is to take into account how listeners hear and remember the speech of others.
"This is particularly important in the context of 'earwitnessing', where a listener has been exposed to the speech of a criminal (such as during an armed robbery, a violent attack, or a threatening phone call). As expert listeners we must ourselves guard against the possibility that auditory illusions will affect the way we hear and transcribe speech, in spite of all our training.
"There is always a risk that we hear sounds which aren't there, or miss sounds which are. Students on our unique MSc in Forensic Speech Science programme are trained in speech analysis techniques which are designed to bypass these problems, among which are computer-based methods which treat speech signals as just another form of sound energy.
"These strategies can be of great help when attempting to resolve the content of noisy or otherwise poor-quality recordings of the type we regularly encounter in forensic casework, and when evaluating the evidence that a suspect recorded in a police interview may be one and the same person recorded committing a criminal offence."