Chances are, youll hear it as laurel too.
The original recording came from Vocabulary.coms 2007 effort to include pronunciations for the sites most commonly looked-up words.
This was the recording made for the wordlaurel.
(Opera singers are trained to readIPA, the pronunciation code that dictionaries also use.)
Jones personally read about 36,000 words for the site.
I asked Tinkler if there was any chance he had the original file for comparison.
And its clear enough that we can now tell exactly what happened when the file was compressed.
What Got Lost
Compressing the laurel clip had a major casualty: the secondspeech formant.
Sound waves each have a frequency.
The higher the frequency, the higher the pitch we hear.
But the sound of speech includes many frequencies at the same time.
Ourearsandbrainspick out the strongest frequencies from this mess of sound.
Speech scientists call theseformants.
The two lowest-pitch formants give us enough information to tell the difference between one vowel sound and another.
But, he conceded, theres just enough ambiguity that you could hear laurel if youreallywanted to.
(He suspected at first that the file was carefully crafted to be ambiguous, an intentionalaudio illusion.)
For the record, I have never heard anything but laurel from this file.
Wavesurfer agrees with my dad, and reads this file as saying yanny.
To understand what happened, take a look at the red and green lines.
These are the first two formants, which well call F1 (red) and F2 (green).
They both hover around the bottom of the spectrogram.
But then look at the processed clip, the one that youve been sharing with your friends.
(poke the arrow in the slideshow above.)
Its noisier overall, thanks to the mp3 compression.
Obliterating the second formant changes how we interpret the sound.
It turns out that phenomenon, too, comes down to the speech formants.
