Luarel vs. Yanny vs. Renoise

Maybe you’ve heard the Laurel vs. Yanny thing? It’s a kind of audio illusion where different people hear different words.

Anyway, here’s an XRNS!

It’s a kind of audio illusion where different people hear different words.

I’ve always called it “being married”. :wink:


if you listen to the sample (I already had ripped it before because of curiosity what the fuck it is 'bout) for long enough, it will sound like some american saying “aural” while some pressed voice high pitch hissing obnoxious alien living in his head will hiss “theory” at the same time, with the sound comming through the american’s nose.

“aural”/“theory”? maybe the sample was crafted to begin with (it definately is strange…), and the sci dude who created it had chosen those words as they were in front of him somewhere because of his profession?

Maybe the difference is three part, first being listening on crap speakers (no bass and mids, accent on certain high freqs…), second being that some people have busted ears (or old age ears, or differences by birth) that let them hear some frequencies stronger/weaker than others, third maybe differences in how the brain processes frequency bands…

I found while toying around with the sample, normally I’d hear what stated in the first paragraph, definately “laurel” and the “theory” being very faint though. if I transpose it down more than 3 semitones it will become something like “yearly” said by some drunk chap who has problems to intonate, and is close to puking at the same time while he is speaking. Maybe “laurel” is just the name of the alien he has in his head instead of his brain, and he thinks laurel ought to get him another beer.

Then I tried maybe the eq balance is the key - so I cut stuff below 800hz, but a peak at 1.5-2.3k and another at 5-7k, slight notch at 3.5k. Now the “yearly” hints through much stronger.

Don’t know if this is accessible worldwide, but the New York Times made a tool that shifts frequencies:

First time I heard the clip (form asome video on Twitter) I played it from my main PC over studio monitors. Heard “yammy”.

Then, same set-up, watched Westworld.

Then, on my laptop (with tiny speakers) played the clip and heard “laurel”. Went back to the PC/monitors, played it again, still heard “laurel”.

Makes me wonder if my ears had become less sensitive to lower frequencies while watching that TV show.

I recalled that every human has a different ear canal resonance frequency, usually in the range of 2-4khz. It is like a boost eq that is also defining the range where the hearing is most sensitive - so maybe different people’s ears emphasise different frequenices. That might be one part of the story - EQing I found that range be most critical part in understanding either word.

@neurogami, I found that after listening loud (yet not numbing…) to high frequencies, my hearing is more acute in that range for a while. So it might not have been the mids being desensitized by the TV, but the highs having been excited and becomming more sensitive. I think the high freq basically hints “yanny” by the upper formants which are very pronounced and boosted in the sample, while the lower freqs/formants (f1 and f2) make it seem “laurel”.

I played around with the shit in renoise, trying my own morph. Only EQ and filters, no repitching! I also found two other “laurel” samples, and the effect will work with them as well, the EQs just need a stronger nudge than with the original…

EDIT: I’ve added a more clean version of my tunings…

Another way to think about it is in terms of undertones, fundamental frequency, and overtones and how the boundaries between those are fluid. It’s also interesting to think of this as taking the fundamental of one sound and substituting the overtone of another.

Applying the band reject filter about about 520 Hz causes me to hear “yanny” quite a bit more.

Also, I think it’s worth observing that there isn’t much of anything above 6500 Hz. Maybe due to compression. The original recording might have been more effective.

I’ve made a more clear version of my eq experiments. was a nice excercise! however a tricky one - I was mixing a moving target. whenever I found it sounded like “yanny”, I went to another sound, came back, and found it was still not quite there! psychoaccoustics in action… Well I circled around the mixing, until I found it was enough waste of time…

seriously, I think it could also be connected to hearing damage at around 1khz. I think I remember it is common - you know, the mean loud painful signal tones, or screechy brakes at the train station…using a notch at around 1k and tuning it carefully makes a lot of the yanny game happen already…

maybe the original sample just happens to hit the sweet spot of casting enough people to misunderstand it, if speakers were sub par or ears are broken. The other 2 samples each needed their own push to go into the “yanny” direction (or “hear’n’sea” with the 3rd, no american speaker it seems).

Another one:

I did some more processing and shit, and think I’ve found out what makes the laurel vs yanny thing. either it is some very fortunate noise accident, or it is deliberately forged, and I think I know how it was done and why. certain things about it (numbers…) are too obvious for it to be not forged. The 2nd voice is not saying “yanny” but “hear me” and is pitched up about 5-7 semitones, probably to fit the technique used. It is mixed into the original sound in a very special way…and that is also why the illusion is failing when it is pitched down, then the “hear me” very quickly will become noticeable to anyone if pitched down - though in a noisy way so you don’t immediately get the point…

so it’s nothing to do with hearing damage, just psychoaccoustics in action. maybe certain parameters of hearing can be different from person to person, so some people might in fact favour hearing “yanny” (after being told that word is in there) over “laurel”. this of course adds a nice confusion to the whole topic. also filtering out the mids/low (cheap speakers, cellphones) and certain frequencies that come by hearing damage, the chance is maybe you will rather understand the payload instead of “laurel”

the original sound, the site was mentioned on reddit:

notice how it is the same voice speaking, but the “yanny” thing is a whole lot more noisy, to conceal what has been done to it. I did not need the original to find out what was going on though.

the “green needle” trick works with 2 videos maybe? expecting the listener would take it he had either heard what he was thought, or by accident the other word he was told, and that the average listener would check at least twice to see if he can understand the other? I just doubt an accoustic illusion being so good. Also the light seems to pulse in sync with the words spoken, maybe that also plays some role.

Please look.

That is what his voice sounds like. If it is his, that is… whoever he is… sound’s like a cheerful old man…

Ya! I love riddles like this…

Another one:

The whole laurel-yanny thing didn’t really convince me, but THAT one!!

Here I dropped my link in the original posting of the sound file on reddit (at least it looks like it is…)

I don’t know a shit about reddit, I just wanted the fun of showing people what I found in that silly thing. Maybe som of you want to upvote?

Edit: hell yeah, if I find muse during the next days, I might try extracting the “hidden message” from the “brainstorm” vid…I fear it is not as easy as with the laurel/yanny thing

Another one:

What the fuck!

“brainstorm” is “laurel” on a bitcrusher, and done much better…

More analysis:

Lots of technical detail.

More analysis:

Lots of technical detail.

I downloaded the mp4 from twitter and copied it into Renoise, and I didn’t see measurable activity above 6500 Hz as I mentioned. The language log analysis displays and discusses frequencies much higher than that, so perhaps it was a recording made with a microphone held to a speaker or something.