Correction: An earlier version of this story described the voice in question as computer-generated. In fact, the recording originated with a human voice.
Not since The Dress has the internet ripped itself in half over a sensory puzzle like this.
In case you haven’t been online in the last 24 hours, that puzzle is a two-syllable sound clip that appeared on Reddit and has since gone viral.
When you hit the Play button, a voice says the word, looping it over and over. The word is “Laurel.”
Or maybe “Yanny” (pronounced “yeah-knee”).
What do you hear?! Yanny or Laurel pic.twitter.com/jvHhCbMc8I
— Cloe Feldman (@CloeCouture) May 15, 2018
Some people hear “Laurel” and will go to their graves insisting that there’s no other possibility. Others hear “Yanny” and nothing will budge their opinions. In-office polls put the perceptions right around 50-50.
I heard “Laurel” on my laptop. My wife, at her office on her phone, texted that she heard “Yanny.” We were each dumbfounded.
When she got home, though, listening on a different device in a different room, we both heard “Yanny.” But when I played a pitch-shifted (higher or lower) version, it was indisputably “Laurel” for both of us.
Ok, so if you pitch-shift it you can hear different things:
— Steve Pomeroy (@xxv) May 15, 2018
In desperation, I sought an expert opinion. Dr. Bradford May, an auditory scientist at Johns Hopkins University, puts it like this:
“Computer synthesis programs can produce unnatural sounds that fall on the boundaries between the two sounds. The listener will place these chimeric sounds into one category or the other depending on the best match. Because humans have differences in their auditory function and category boundaries, some will hear ‘Yanny,’ while others will hear ‘Laurel.’ This relates to the difficulty some Japanese speakers have with R vs. L sounds, which are not distinguished in the Japanese language.
“Humans learn language by learning to attend to meaningful sound patterns, and by learning to ignore specific acoustic features of those sounds that are not important. So, a child, an adult female, and an adult male will produce very different acoustic patterns when they say ‘Yanny,’ but you will hear ‘Yanny’ because you are attending to the underlying meaningful pattern, not the talker-specific acoustic features. This is called categorical perception.
“You will never confuse ‘Yanny’ and ‘Laurel’ when spoken by actual talkers, because they can only produce natural sounds that fall within the distinct parameter spaces of the two sounds.”
The origin story
The original sound, by the way, is indeed the word “laurel.” You can listen to the original, right here—it’s the human-recorded pronunciation from the word “laurel” on Vocabulary.com. It was captured by a teenager who was listening to the playback of a professional pronouncer’s voice over her laptop speakers, then posted to Instagram, from there to Reddit, and the rest is history.
Marc Tinkler, president and CTO of Vocabulary.com, says that in 2007, he hired professional singers to record the pronunciations of 200,000 words. (Why singers? “We had the words, plus we had the IPA spellings of the words—International Phonetic Alphabet,” he told me. “It’s basically an alphabet that describes every sound, every phoneme, that you can make. Opera singers study IPA, because they need to be able to sing in different languages.”)
Tinkler notes that even the original recording, played off his own site, sounds like “Yanny” to some Vocabulary.com employees—at least when played “on a TV we have in a conference room that has really terrible speakers.” Tinny speakers, like the ones on TV sets and laptops, tend to emphasize the higher frequencies, which brings out the “Yanny.” Older people, who have begun losing some of the higher frequencies in their hearing, are more likely to hear “laurel.”
(Tinkler also says that the Yanny/Laurel craze has done a real number on Vocabulary.com’s traffic. “‘Laurel’ is far surpassing any other word,” he says. “We’re looking at three, four, five thousand people on the page at any given moment.”)
The bottom line is that we hear different things because our hearing is different, and because our brains categorize differently. If you can hear only one word or another, this will help to open your mind: The Times made a slider that lets you adjust the frequencies so that one word or the other jumps out.
As with The Dress, someone has stumbled upon an absolutely incredible razor’s-edge bit of hybrid sensory input. It doesn’t matter if you hear “Yanny” or “Laurel”; the good news is that you’re right.
David Pogue, tech columnist for Yahoo Finance, welcomes non-toxic comments in the Comments below. On the web, he’s davidpogue.com. On Twitter, he’s @pogue. On email, he’s email@example.com. You can sign up to get his stuff by email, here.
More from David Pogue: