FreddieMeter - Queen + voice recognition bullshit

Here's a little singing contest variation dedicated to one and only Freddie Mercury. It is called "FreddieMeter" because you know - it meters the Freddieness.

Why? Because he was sooooo great, you know? And he had such a great voice, do you know about that? Can you sing like Freddie Mercury? How close is your timbre and melody? Let's find out!

Give me a break.

While Freddie Mercury was great-ish (probably) and stuff (for sure), i don't think this posthumous celebration of his talent is all that necessary. The man has a body of work and it is literally everywhere and if you want to celebrate him - just listen to his music and stop looking for a reason to shove it down the throats under some new pretense via marketing effort. It doesn't add anything substantial to the conversation

Seriously, it is dumb experience marketing gimmick cringe. You take a hyped-up topic like Machine Learning and combine it with mildly promising idea of voice comparison. Then you need to make it broadly available so you regress it to the simplest form. And then you tack your brand on it so that it would "matter" more and generate that special kind of "engagement".

The reality is a bit different. Here's what FreddieMeter really is - it is a throwaway piece used to momentarily spike the interest in the brand and then go away without notice because no one will care about it beyond the inital splash. There is just nothingness all around.

Also - with this service you can compare Freddie's voice with Paul Rodgers via some audio input tweaks. Paul Rodgers is cool too (and he's probably a Chtulhu who got Doug-Quaid'd with some plastic surgery, hypnosis and Revolver Ocelot-level of mindrewiring), and did a Queen album too. It is probably the most useful thing you can do with it.

The other part of the problem is that it was created by the Google aka "the big tech company". Which is a kind of a red flag, you know? The thing about big tech is that they are infovore - they feed on information, and they want their information diverse and they will go a distance to get it. Case in point - this thing. It is a ploy to get more data. And brands go for it because of "whatever works" mentality.

I mean - this service will likely generate some audio-visual data of people singing or lipsyncing. This data will likely go down into some facial and voice recognition and transcription research stuff - and at some point down the line it will be commodified as some service.

The fact that there is a statement "Your audio doesn’t get uploaded to servers to be analyzed, so your vocals stay private" doesn't really mean anything - you need to know what to check in order to find out whether it is really true, and most of the people just don't care and follow the impulse. It is foolproof because it seems harmless.

Then there is a problem of purpose - what kind of endgame this project expects? What is the end goal? Just to have some fun and "celebrate the icon"? So, you can record your voice singing some Queen songs and compare with the way Freddie Mercury sang it. And that's it. This use case is flimsy to say the least.

What kind of insight does it really offer? That your voice is more or less reminiscent of Freddie Mercury's according to some broad criteria and lots of hidden parameters? OK, what is the benefit of having this kind of information? What it is good for? It's not a white elephant artistic technical achievement, so why bother spending resources on that? Because Queen brand paid for it so that they can raise awareness and celebrate? Do you really need to do it that way? There are so many questions left and somebody got paid for this.

The idea itself is fine. But that's not how you can use it and find out something interesting. One of the problems with these kinds of projects is that their mechanics allows much more than the business need requires. Which makes the case for the thing known as "the waste of effort". And this begs the question "why bother?". And the answer is - 'cause it generates data that can be used elsewhere. Period.

But how you can use it more productively?

One of the ways this idea can be applied for the common good is language learning. Speaking practice is one of the integral elements of the process and it is not really available for everyone. This kind of service can bridge gap and help people learn foreign languages in a less complicated way.

On the other hand, this kind of approach can be used to study the chrological changes of the way people speak througout the years and time periods. You can trace the evolution of speech patterns, tone, timbre of the voice of a specific individual or a group or community. And that is going to tell you a lot of things about how the way of verbal communication permutates and adapts to the surrounding world.

You can also singalong with Chuck Shuldiner. He's dead too and deserves to be celebrated.

How OpenAI sells GPT-2 as NLP killer app?

So, the thing we've been waiting for a better half of the year just happened - OpenAI had released a full unabridged version of their infamous Natural Language Processing model GPT-2. 

You know, the one known for being capable of generating texts that look and read just like the ones written by actual human beings. The one OpenAI deemed to be "too dangerous" for the public release. The "Pandora's Box" that, if misused, would probably flood the world with fake news and otherwise manipulative synthetic content. The one that would make regular copywriters and journalists utterly obsolete. The one that birthed AI Weirdness blog, after all. You know, that one. 

Well, it is back again and it is a good reason to talk about one of its lesser-known aspects. 

One of OpenAI's competitive advantages is that it is "capped-profit" limited partnership, which means that OpenAI doesn't have to generate profit per se, but their sponsors (in limited partnership - folks like Microsoft) can use OpenAI's research for their own purposes (products or solutions within products). At the same time, OpenAI's primary performance indicators are more or less spread of their technological solution and maintaining its notoriety. 

Let's talk about the latter. 

The thing you need to know about OpenAI GPT-2 is that its marketing campaign was pitch-perfect. This is how you market the technological product to the unwitting masses. This is how you present a value proposition to those in the know. 

The secret is that you mostly don't really talk about the product itself but instead concentrate on its impending impact, the way it will shake things up, "make a difference," and so on. Hence "too dangerous" narrative. 

This approach creates a mild case of hysteria perpetuated by an eerie mystery, which is mostly based on a bunch of "um"'s and "uh"'s and whole lotta sweet technobabble. The goal is to exemplify what the thing is capable of subtly - leaving enough space for speculation. On the one hand, this approach created a stir in the media, thus guaranteed a higher degree of visibility than competing products. On the other hand, it signaled to the right people that there is a multi-faceted solution waiting for proper implementation.

Just look at the launch sequence and how it establishes the product. It kicked off with the press release that stated, "oh well, we guess we just made some kind of doomsday device. And we are really concerned about it. And we're definitely not going to do anything with it, because it would be ridiculous to put it out in the open just like that. So we are just going to keep under wraps for now". 

What happened next is the tortsunadomi of news pieces that retold the story over and over again and also added a few meandering thoughts full of piss and vinegar to the conversation. It is then quickly incorporated into the "fake news" and "deep fake" narratives and firmly contextualized as something definitely "threatening." As a result, the burning mystery is perpetuated and elaborated. 

The next step was to provide a proper showcase of the model capabilities while avoiding spoiling the big thing. It is a great sleight of hand. The whole thing is out in the open, but you claim it isn't. The showcase is made by an abridged version of the product presented as a technical paper and a tool for researchers. 

It played out perfectly - the presentation of expanded and diversified by multiple points of view on the model. As a result, - more opportunities became apparent.

There were a couple of NLP projects that provided GPT-2 a lot of publicity. For example, Talk to Transformer presented the original use case of GPT-2 - text prediction. The way it was presented neutralized any emerging criticism regarding its clumsiness (for example, it couldn't predict Austin 3:16). It was simply freestyling a bunch of stuff - sometimes it was more coherent, sometimes it wasn't. 

Then there was the Giant Language Testing Room that visualized the text analytics and showed the mechanics behind computer-generated and human-generated texts. And then there was Allen Institute's Grover than was designed to expose deepfake texts and also showcase how easily the model can recreate the text with a couple of settings tweaked (except when you ask it to write an essay on Tawny Kitaen for some reason). 

Also, there were numerous blogs that applied GPT-2 for creative purposes, like writing unicorn fiction or generating incongruous names. Blogs like AI Weirdness that basically celebrated model's bumbling nature.

In a way, this kind of presentation operates on a similar framework to Marvel Comics' built-up the arrival of Galactus back in 1966. First, The Watcher shows up and warns about the coming "doom." And everybody go "ah, we've gotta do something!" Then you get a harbinger in the form of Silver Surfer, who wreaks havoc but at the same time continually reminds us that "the worst is yet to come." The heat is right around the corner now. 

And then we get the main event - Galactus, who was just another supervillain in the long line of "ultimate threats" just ten times bigger and because of that characterized as ten times more imposing and menacing and dangerous because authors of the comic said so. 

And that's what happened. After months of build-up and speculations, in early November, the big 40GB 1,5 billion parameters version of GPT-2 was released to the public. And it was exactly what it said on the tin from the very start, just ten times bigger. However, the momentum generated by the build-up made it seem less apparent. But there is a catch which makes things even more interesting.

The thing is that "ten times bigger" doesn't translate into "ten times better" because the technology doesn't really work that way. What matters is how it manages to accomplish specific use cases and how does it correlate with the solutions for actual practical needs. 

GPT-2 is presented as a tool that generates texts that seem natural as if an actual human being wrote it. It is cool, but you need to understand that the model operates on the finite dataset, and all it really does is derives the text out of probabilities and context approximations. Text generation is not really generation. It is more of regurgitation in sophisticated combinations. GPT-2 output looks and reads just like a real thing in a way that the real thing can be bland and passable and not really saying anything, just like loads and loads of content on the internet, especially, its corporate parts. 

What GPT-2 really does is creating a serviceable text product that serves a specific well-defined function. But you can't really market that because it is not really unappealing beyond those in the know (say hi to BERT). However, that's the absolute peak of practical NLP. 

A perfect example is a conversational interface. You say "hi," and it says, "hello, how are you?" because that how conversations usually go. The other example is a verbal interpretation of data like stats in Mailchimp where you can get reports like "22% opened your newsletter, 5% of which clicked on the link, and 1% gave it a thumbs up", but it is beyond current GPT-2 use case. 

That's why the "fake news" narrative is so beneficial for GPT-2. 

On the surface, fake news seems to be a natural avenue for that kind of thing. That game is all about quantity, not quality, and the NLP generator can help with that. It was hammered over and over again in the press releases and subsequent news pieces. And the use case seems legit. Technically, the goal is to generate derivative content that will reinforce and possibly perpetuate prerequisite beliefs of the target audience. Just what GPT-2 is good at.

Except, that's not really the message of the narrative. The real message all along was, "this thing can be used like that if YOU won't use it." And that leaves an intriguing mystery of what is going to happen next, which in turn engages the target audience more than the product itself.  

Count off, sampling and Kirk Pearson's unpersoned "Going Up"

Last Sunday, I was talking with my friend about De La Soul intense sampling style and at some point the conversation veered further into conceptual art territory of supercuts and Youtube Poop. In particular, I was asked whether there were any pieces that combined counting and sampling that weren't DJ Shadow's "The Number Song". And I remembered one - i saw it a few years back. I just couldn't remember its title.

So i've spent over 5 hours trying to find this particular plunderphonic mashup. What a nice way of spending Sunday evening. For some reason it was way harder than i thought. 

Here's what i remembered: the piece featured a countdown from one to one hundred culled together from various songs. Plain, simple and to the point. The best supercuts are always like that - they just blast through and leave you contemplating.

I first saw this piece mentioned in one of the Rate Your Music lists back in the early 2010s. After that, i saw it featured on one of the topical wikipedia pages, either Sampling or Mashup. I also vaguely remember it being on Free Music Archive. The only things i couldn't pin down was who made it and what was the title.

At first, i checked rate your music and couldn't find anything resembling that particular piece. But i have found a list with many-many unusual types of count off in pop songs. It was insightful and ultmately pointless.

Then i have checked whether the piece was still referenced on the aforementioned "sampling" and "mashup" wikipedia pages. None. I also checked "plunderphonics" page. If there were any mentions of this composition - they were long gone. Or they were still there and i just couldn't identify it correctly due to my recollection of the piece being incorrect or misrepresentative.

Then i went further down the wiki wormhole and checked pages like "Count Off" and "Counting (Music)". Once again - nothing. 

Then i went all guns blazing on google and typed every imaginable description of the composition i could think of - "plunderphonic counting off/count off/countdown composition", "countdown out of samples from different songs", "count off sound collage" and so on. There were three types of search results: 1 - DJ Shadow "The Number Song" (obviously); 2 - every children counting song ever (predictable outcome); 3 - general references to respective concepts (because there was nothing else relevant).

It felt really awkward - i was certain that i heard that piece but i just could not find any trace of it. At some point i even thought that it was some kind of Mandela effect thing or that i just imagined that piece and then forgot about it and now thought about it again but as something else. It was really confusing.

Then it hit me - if i can't find anything on the current web - why not peek at the past? After a thorough raid on Wayback Machine on every available lead - i was able to identify the track.

The name of the composition was "Going Up". It was made by Kirk Pearson. The composition was released in 2011 on an album "Please Don't Sue Me". "Going up" is made out of unaltered soundbite samples from different songs that together create a count from one to ninety-nine. The selection of samples is eclectic - among other things you get obvious Doors reference and less obvious Zager and Evans reference.

Here's what is interesting - it is scrubbed from existence. I've looked through every available Kirk Pearson release and couldn't find it. Not even the slightest mention. There were lots of stuff but not "Going Up". Why? Guess what - there is only a few things that can make things un-exist on the internet. In case creative works - it is mostly commonly a copyright violation.

The thing is - the whole "Please Don't Sue Me" album was a giant lament configuration of plunderphonics - it was a detournement of other people's stuff - culture jamming in action. The thing that propels the culture forwards and opens up new possibilities and explores new aesthetics. 

And naturally at some point the representatives of these people took action. Because we can't have conceptual art that transforms other pieces and recontextualizes them to explore certain themes and ideas - it is disrespectful and detrimental to the sources. 

It is very sad, but i'm glad i took this journey. It made me think about how unsafe and fleeting is culture jamming. But at the same, it encouraged me to move on. It is an act of defiance, and that's what art is all about.

Now, let's perpetuate the legend of "Going Up".

Funny thing with the cover Hail to the Thief

Back in mid-00s i was trying to unsolve mystery - why people like Radiohead? I just couldn't get it - they were too boring to me. And still are. I find their music to be full of interesting ideas buried under nonplus performances and generic songs. That's my opinion. You don't have to agree.

The only thing i really liked about them was their visual presentation from 2000 to 2003-04, in particular the cover for their 2003 album Hail to the Thief. (I will write about Kid A blips someday). There was just something endearing about. It promised something else. 

Made by Stanley Donwood, the cover was a roadmap of sorts covered in advertising buzzwords mashed together in colorful blocks.

If you separate the cover from the album and look at it as a thing of its own - it is pretty good piece of conceptual writing.
This cover is all about plunder and recontextualization, hijacking of aesthetics and making it interesting again. While it definitely tried to replicate and multiply the vibe of Ed Ruscha spliced with opaque statements of Lawrence Wiener - it works as a statement on commodification of language.

The words are stacked in several columns, they represent commercial products and concepts in no particular order - just a whole bunch of stuff. It is kinda like listicles Roger Waters used to do in the latter days of Pink Floyd.

The sheer volume and density of content make it an open-ended piece. You can read it any way you like and construct your image sequence of the moment.

***
But i did something else. I like to OCR things that are usually not fit for OCR. My original intention was to shortcut typing down the words so that i could make some kind of a generative piece. However, the result was exceeded that.

Here's what i have got instead:

RE a att Peet ta
Ae 8 Pat CR ss ree eee, ai : Sekt rd Ceo a 4 lp ae Tak SO Pa > bh cane? ae wees ott, 7 eS et ha *
Bg Ue eee ok EEE SEE PEE ae a eal Te Aaa] ysoes 4 Peete eee Oa | oh, Pay he ea eae EE pe ge Ad Sa a EA BE
std eg shee a ee cosh SD deo Sie ie ie 4 ef i HA Ss
bee GS a a wfo* eee Ly eae 1 oe eas ee gh EAT 2g Bee we A ek aes Pe OD Fae ee BERS PEPE P, wpe Ps he BSS
Be get ai tie! aie La VEEL ES ostaen aaeee es aoe aoe Ba AU ki Up coe Oe
PS ee Oe Bee Ee up SE Regt eye | see Signe! gabe Aeg ASCos
bo La) gery: “pee CHR peers Scitee pee heey sil epREE eR Sas ce SMR AE
ett Tee Stay ee Be 1 PEE woe i MOE pee Wee as aN aoe ‘ SOG REGS in . SY
POR AS Ege gre ee ea ot OPA Cae Bo teeter re Gof res ES ae he Ss as BS
eR ere ¥ | ob eae ee Be Ry eee a KO ees
es Ea Ee ES gO ee 8 VOR SARC RRERN iel e te at AY net eS a Sy
oD RI By AER Be a AHL Blo” 33 Set alee SB Oo "Up REE SN
Ce AE a eee IPOS ge Ns I BN Ae ee Aer ae AL PSE Spe hee ee SRE
eR fa See ae Be: Pity eee INE Mt a i i PSR ee pee iT SEE EIGEN SS Gi
ae ea”. a ele ee ee i Vat A ISAS LE us Pf AO ay ew UNS
OE Mi EE ee kN 2 is { Sef Ps BE = Piacoa pi Nad cd: Derg
i Lp! BG oe ae ee ee PRE ie =
ui fat it aa per, 2 Mee 3 ar Cova) A if fee ‘ . as £ : cre ae uid Fe at “¥ me F Wy it oe: ~af 2 4 2 . 88 as Se PENG Cote SS .-
ba Petia hee OAD a Me i Ein RET
di ge A ot ce isk MOE Lcd eae | ir ee 7 EAD Spee oY, we a eS
LEE Aa ean BPEEEE ac illaacaiatel Li ; ™ Bf ed | RAE SS SSG
(he Pe SANS %e Were ABS ES
GP | BP CB SEE ARS ee ‘ wy DP ka BEARS PR
oN Ady ae A ae . iW | Oe AW SS
idf oF page eee Zot > Pa Wa : a AN ie AER
Sg oe ty f a 7 a “Ws, >. WS he ass
ae) Ve Pps a | Fai WAP LS
af pte f | GIRLS H 0 = Vy he MA
shi aoe a , s M eae a top
a GASOLINE [77 E Uy
Pai OCUN Bs alg GRAND Br Tye
g Pate eaaae SUICNO| fm : BP iy SOL Pi
c ; ; S a : | Ni ~~] i PPANGER 3 . ze A | ) t f lt
Po an Dé ys
PATROL] ANYTIME Mfote rage ND Bee eA Ne opr
LUBE. GOD ! TRUST, v Tos OTT COEF 7 & Bis
OR LTGP Re Shs
= 7 ior awtls AAR
VCR iiine ses
Mri BAUD SRRS
A\e | WIS SFY
LOS$ LIPS SAM
a yi = Pw
sale Sf THEFT Dore, NT, TSS
CONTAIS DTN SANE poo Aas apes, MEDIA
4 ee i . , ‘FE y 4
c . ay; uh ma j Ey XTRA Ch PO° SPIRITUAL
OSED ea ati COlF :
Am, | mee §!QU0 UNAREMEMBERIN G
INTERNET Mec) 0 — PGR HAMBURGER
a OME Re OS Be ol al
USED Mee | ARITA
a ANT Bai;
Pee ee a hm Re ik
‘ey etek
F re:

***
One of the curious things that happens with OCR is that it requires keeping the input images straight in order to do the job right. Even the slightest presence of an artifact throws a wrench into the operation. The more artifacts or otherwise odd document features - less accurate the result.
That's what happened with the cover of Hail to the Thief.

There are bits and pieces of the original words sprinked here and there but the majority of words are mangled, sliced, chopped and screwed. At times, it looks like computer-generated madlib. Other parts looks like they were transcribed from a backwards speech.

But it made me think about the texts hidden underneath the texts on the images, the ones only an OCR algorithm can extract. I know why it happens that way, it is a side effect of the algorithm specification. But what does it means from the conceptual point of view?

Money is not our god...

More of the same song and dance. I've been talking with a friend of mine who is project manager about the role of poetry in life.

She said it is whole lotta bollocks rolling in the back end. I think she's right even though she doesn't get it.

The thing - it is all about breaking through, fighting the conditioning of the establishment.
Of course it is bullshit. how else would pure distillation of language be perceived by grounded and continuously pounded folks who operate on templates and act on cut and paste, who are numbed out to the point of incompatability with the thing? Seriously, would you expect anything else?
And then she said "well, i'm making more than all of your poets combined". Yeah, for sure. Let me tell you something.

Back in the day, i was paid for an article in low nominal banknotes - ones and twos and some fives. There were like two hundred of them - really big pack. and most of them were worn out real bad.
Some of them were rejected at stores. It was frustrating. And then i remembered about spocking technique, took a marker and wrote some texts on these wornout bills. Because what else you can do?
The early ones were right in the "obey" alley. But it got old on the third bill. Quoting Barbara Krueger in ukrainian is like brushing teeth - nothing special, except it still stinks pretence.Then i went for zen koans, it was fun while it lasted. after i ran out of wise-sounding words, i've dropped some random words and exclamations. It was some henry rollins reads ee cummings stuff - vagabond weasel kazoo abbattoir and so on.

After i was done - i had a hundred or so defaced bills. It looked like a brick going undercover and pretending to be a piece of sliced bread.

I stapled it together and dropped at the nearby coffeeshop. I've bought a largest cup of tea imaginable and went to the corner. The reaction was weird.

A group of hipsters decided to find who did it because it was disrespectful. They were freaking out because of some barely intact defaced cash. You couldn't even buy a good coffee for that sum in that coffeeshop. But they were furious.

They attracted a lot of people to the thing - they were showing what an atrocity it was, how tasteless and witless it was and people were like yeah whatever moving on. After a while they cooled off and started to feel awkward. They couldn't do anything with those bills. Eventually, they left it alone. But it took them like three hours and they looked shellshocked .

It felt like victory.