Would you like to be a blob?
Now if your answer is already yes, I am thoroughly curious as to why, but in the highly likely scenario that the answer is no, I’ll sweeten the deal: would you like to be a blob that also happens to sing opera?
Still no?
A shame… but let me try to convince you anyway by introducing you to the wonderful and thrilling world of Blob Opera created by David Li in collaboration with Google Arts & Culture.
Unleashing the power of the blob
If you click on the link, you will immediately gaze into the bulbous eyes of a little multi-coloured blob-like creature and be given the prompt to basically squash and stretch said blob. To what end? Why, if you comply with the instructions it becomes crystal clear in a matter of milliseconds: to make music of course! You see, the more you stretch a blob the higher the note it sings will be. The more you squash it in on itself… well, you get the gist of it.
Once this basic tutorial is completed, Blob Opera will show you that you can not only control one blob, but four simultaneously. While controlling more blobs is always a good thing, here it has a practical use. In addition to your initial bass blob, you now have another three voices at your disposal: a tenor, a mezzo-soprano, and a soprano. Making them work together is very simple, if blobs are lower-pitched than the blob you’ve selected, they will follow the lead of the selected blob. For example, if you mess around with the mezzo-soprano, the tenor and bass will match the mezzo-soprano’s notes.
If you want a blob to not sing for a period of time, that’s possible too. Simply use the mute button and you won’t hear a peep from it. You can also decide to change it up a little, as sliding the blobs to the left or right changes the vowel they’re singing.
Here’s a little taste of what can be done.
Inner workings of the blob
Okay… but what’s so special about it?
Other than being hilarious, Blob Opera is actually the product of Machine Learning (ML). In other words, the different notes that the blobs are chanting have never been sung by anything else than these blobs.
But how does that work?
The crux is that ML has a very different way of analysing data than another algorithm: instead of drawing its knowledge from a fixed database ML is capable of drawing conclusions from the information it has been fed. ML can thus produce new information, in contradiction to a traditional algorithm that is only capable of behaving within the inputs and limits its creator has established for it. In the words of the father ML, Arthur Samuel, ML is a
‘’Field of study that gives computers the ability to learn without being explicitly programmed.’’
Arthur Samuel 1959
For Blob Opera, the ML used has been fed sixteen hours worth of recordings of four different opera singers who, you guessed it, correspond to the four different blob voices. From that, the machine deduced what sounds corresponded to different notes and vowels to finally give you the magnificent blob rendition. It is no wonder then, that this voice rendition sounds way more natural than speech synthesis (also known as text-to-speech) as the ML doesn’t need to pick and pronounce every phoneme individually.
The blob’s place in society
Now more than a fun way to interact with ML, this endeavour also has a more practical aim: transform the skill that is opera singing, into a tool usable by all. This raises some interesting questions for the future of arts, because if these kinds of tools are perfected, do we really need to put effort into learning the craft at all? If everything is but a click away…
Photo by Ali Shah Lakhani on Unsplash
Question for another time perhaps as, as it stands now, there are still limitations within the ML art world. In the meantime however, nothing stops us from having a little fun with the blob choir.
I saw these blob on tiktok. I find it so funny that you are writing about it!!! hilarious