The Last Recording Artist

A comprehensive guide to ghostwriters, virtual pop stars, and the world to come.

Recently, the New York Times published an article about a powerful new piece of software with the potential to change the music industry forever. Written by former Billboard editorial director Bill Werde, it describes a tool that can take existing recordings of a particular singer’s voice and synthesize them into “new” performances by that singer. It can do this without their direct involvement. “Imagine having a singer with a world-class voice at your disposal, any hour of any day,” the piece begins. “She’s just standing at the ready, game to perform whatever silly song you might make up for her: a ballad about her love for you, a tribute to your best friend’s golf game, a stirring rendition of the evening’s dinner menu.” He goes on to suggest that “close friends of Madonna and Mariah may already have had that pleasure” but that for “everyone else,” this new technology offers “the next best thing.”

Developed at Pompeu Fabra University in Spain and financed by the Yamaha Corporation, the software, which is due to be released to consumers in January, allows users to cast their own (or anyone else's) songs in a disembodied but exceedingly life-like concert-quality voice. Just as a synthesizer might be programmed to play a series of notes like a violin one time and then like a tuba the next, a computer equipped with Vocaloid will be able to ''sing'' whatever combination of notes and words a user feeds it. The first generation of the software will be available for $200. But its arrival raises the prospect of a time when anyone with a laptop will be able to repurpose any singer's voice or even bring long-gone virtuosos back to life.

By “recently,” of course, I mean twenty years ago. Vocaloid was developed by Kenmochi Hideki and his Barcelona-based team at the turn of the millennium, and Werde’s New York Times piece was published just a few months before the first commercial release of Vocaloid software in January of two thousand four. “If Napster and other online file-trading programs have taught the world anything, it's that once a technological cat is out of the bag, it can be difficult to control,” he wrote. “What's to stop dilettantes from creating their own fonts? Could it be long before falsified but entirely convincing clips of Britney Spears begging for Justin's forgiveness circulate on the Web—to say nothing of George Bush conspiring with Tony Blair about weapons of mass destruction?”

At first, Vocaloid software was meant to serve as a practical studio tool for producers. Werde called several music industry professionals for commentary about Vocaloid in his piece, and some of the most positive feedback came from The Matrix, a pop production team known for hits like Avril Lavinge’s “Complicated” and Hillary Duff’s “So Yesterday.” This makes sense. The aspiring young pop artists that Y2K-era producers like The Matrix worked with tended to be child actors imported from other industries that didn’t require them to develop basic musical chops. “As producers, you run into some artists and oh god, it's so hard to get the right vocal,” Matrix member Scott Spock told Werde. “It's intriguing, this idea of [saying] OK, just give me all your vowels and all your consonants and I'll see you later.”

The first commercially available voicebanks were pitched along these lines. Called “Leon” and “Lola,” they were developed for the English-language marketplace in partnership with a UK company called Zero-G Limited. At that point, Zero-G was primarily known for selling sample CDs full of royalty-free breakbeat loops to drum and bass producers. That is to say, they were already in the business of selling shortcuts. Zero-G hired two professional vocalists to spend multiple five-hour singing days in a recording studio building up a library of recordings to draw from. The hope was to create voicebanks with “generic soul-singing voices” that could provide producers with a cheap, digital alternative to paid session vocalists. An inexhaustible font of cybernetic scab labor for the new millennium.

All told, this was a relatively modest aspiration with modest implications. Nothing too much more extreme than what happened when the invention of multi-track recording allowed Ray Charles to overdub his own backing vocals during the fifties when he didn’t feel like calling in the Raelettes. Consequently, Leon and Lola would only ever experience modest success. They hadn’t even hit the market yet when Werde wrote up Vocaloid for the Times, and already he was glossing right over them in order to focus on the spicier implications of the technology. The one he seemed most fixated on was the notion that Vocaloid could resurrect Elvis Presley.

Elvis seems like an obvious candidate for vocal reanimation. Recently (and for the first time), his estate licensed a couple of his songs for dance-floor remixes; one of them became a No. 1 single in England. Licensing Elvis for Vocaloid would be a different matter, though, says Gary Hovey, vice-president of entertainment for Elvis Presley Enterprises. ''If someone came to us and said, 'We want Elvis to sing this new song,' we'd have a lot to contemplate,'' he said. ''We tried to retain the integrity of his original song with the remixes. Now you're talking about a whole new vocal performance of a song he never sang or knew? How do we know he'd want to sing it?''

''Believe me, that would go all the way to Lisa,'' he added, referring to Elvis's daughter, Lisa Marie Presley, who owns Elvis's estate.

Still, there is the potential for enormous money to be made, even by Elvis standards. How much would an advertiser pay to have Elvis sing a new jingle? How easily would a new ''Elvis'' song climb the pop charts -- if only for the novelty value? [Zero-G Limited founder Ed Stratton] is optimistic about the prospect. ''No font comes out of the box with a singer's timing and expressions,'' he said. ''It's just the tone of his voice and his pronunciations. The finer bits of expression -- timing, pitch bend, the sorts of things that add real character -- would have to be added by the user working with the font. It would take a great deal of effort to make it sound just like Elvis. But you could do it.''

Stratton was likely just being polite when he indulged Werde’s line of inquiry here, but what he said is accurate. Vocaloid software is specialized and technical. The sky is the limit in terms of what’s possible when the tools are in capable hands, but using first-generation Vocaloid software to generate “new” Elvis Presley vocals would have required both a virtuosic talent for operating the software itself as well as a precise knowledge of Elvis’ vocal style and speech patterns. The necessary expertise would have needed to go far beyond anything Presley would have ever required from actual collaborators during his life. No rational person would ever undertake the amount of hard work required to perform such a specialized task without a guaranteed reward.

Antares Auto-Tune, another paradigm-shifting audio innovation from the Y2K era, achieved mass adoption because it was essentially plug-and-play. People of all ages and technical backgrounds could instantly understand what Auto-Tune is and how to use it. A wide variety of artists from all manner of different backgrounds and musical traditions started to experiment with Auto-Tune almost immediately, because using the program was fun and easy. Operating Vocaloid, by comparison, is like learning to pilot a submarine. It requires an enormous amount of practice and specific expertise to utilize the software effectively. As such, we didn’t see a mainstream inflection point for Vocaloid in America like Cher’s “Believe” was for Auto-Tune. The idea that this software can be used to generate “new” performances by established celebrities has remained unrealized for decades. That is, at least, until last month.

This is a picture of Hatsune Miku. You’ve probably seen her before. Videos of her concerts, where a “holographic” representation of the character performs alongside human backing musicians, started going viral on western social media platforms about ten years ago. She is very popular in Japan, and has become part of the global pop culture firmament through appearances in video games and fashion campaigns like this one from twenty sixteen where she is “dressed” by Ricardo Tisci in Givenchy couture. Most people who are aware of Miku likely think of her simply as a “virtual pop star,” which is more or less accurate. Still, I think it’s worth taking a closer look at the nuts and bolts of what Hatsune Miku actually is. To do this, we must begin by asking the obvious question: who owns this, and how is it making them money?

Unlike Leon and Lola, the first Japanese-language Vocaloid voicebanks were developed by the Yamaha corporation internally. Given the names “Meiko” and “Kaito,” these voicebanks were released to consumers with the help of a Japanese company called Crypton Future Limited. Sales were meager, but Crypton CEO Hiroyuki Itoh didn’t want to give up on the technology. His company began developing a new voicebank, one that was intended to serve a very different purpose than any of the Vocaloid products that had been released prior. A Wired magazine profile from twenty fourteen explains:

In 2007 Crypton’s CEO, Hiroyuki Itoh, was looking for a way to market a virtual voice program he’d developed using Yamaha’s Vocaloid 2 technology. Vocaloid’s first version had not sold well because, it was suggested to Itoh, it didn’t sound very lifelike. He disagreed. (Japan’s appetite for all things humanoid is insatiable, he knew, if properly plied.) What Vocaloid needed, Itoh believed, was an aidoru, an idol. So he engaged an illustrator of graphic novels in Tokyo who goes by the single name Kei. Itoh told Kei he wanted something cute but also slightly edgy, something that would attract creative young people to Vocaloid. Kei came back with a rendering of a 16-year-old girl who was 5’2″ and weighed 92 pounds. She had long, thin legs, coquettish bug-eyes, pigtailed blue locks that reached almost to the ground, and a computer module on her forearm. Her first name, Miku, meant “future”; her surname, Hatsune, “first sound.”

“Hatsune Miku” is the name of a voicebank, but also the character Crypton invented to personify it. While earlier voicebanks were modeled after professional musicians, a voice actress named Saki Fujita was enlisted to serve as the basis for this one. Her relevant experience involved playing fictional characters in animated films, which tells us much about what made Miku so different from her predecessors. Itoh hoped that Miku, like many of the other anthropomorphic mascot characters that exist in Japanese pop culture, would develop her own passionate fan base. If this happened, he believed creators would become highly motivated to feature Miku in their own works in order to expose them to her audience. Miku’s self-replicating fame would then provide Crypton with innumerable opportunities to make money licensing her image and selling official merchandise.

Like Miku’s fans, Itoh has difficulty defining what category of thing Miku falls into. Around the office, he says, she’s not referred as an idol or a character or a cartoon.

“She’s … Hatsune Miku,” he says.

Has he ever imagined a backstory for her? A home, a family, a life before Vocaloid?

“No,” he says, as though the pointlessness of that should be obvious. “Just age, height, weight—and outfits.”

In fact, it would be pointless; Itoh knew that if Miku caught on, her followers would write her story. Such is the genius, beauty, and grotesque fetishistic wonder of otaku culture. These geeky fans are devoted to characters, not Western-style celebrities. Human stars burn out quickly in Japan, but beloved characters—from Hello Kitty to Gundam—last for years. When a character from an anime series or a manga comic, a videogame, a toy line, or even porn catches on, fans engage with it by making new iterations and variations of their own—homemade videos, manga, games, more porn. The Japanese term for this, niji sousaku, translates as “secondary creativity.”

Ten years ago, Itoh had difficulty finding the right word to describe what Miku actually is. Today, the correct terminology is blindingly obvious. Hatsune Miku is a platform in the shape of a teenage girl. She is a means through which the niji sousaku of fans can replace the “primary” creativity of individual artists as the engine driving profits in the world of mainstream pop. She is not a digital facsimile of record industry superstars like Elvis Presley. She is something entirely different. Miku’s business model is a new development, one that is inseperable from the Vocaloid technology she emerged from. The onerous, specialized nature of that software has so far prevented others from replicating her success. That is, at least, until last month.

To understand what it looks like when technology changes music, we need only look at the nineteen sixites. That’s when talk shows, variety shows, and newscasts—most radio stations’ bread and butter up until that point—migrated to the new visual medium of television. This left radio stations with hours of empty airtime to fill, and an obvious incentive not to compete directly with TV by running the same sorts of shows. It was a bind, and innovative thinking was required to discern a away out of it.

The exploding popularity of vinyl records and high fidelity turntables provided a novel solution to the problem. Vinyl discs were much more durable than the shellac records they replaced in the market, which allowed radio stations to start building deep music libraries without fear that the discs would become worthless after minimal use. So-called “disc jockeys” were hired to program long, uninterrupted blocks of recorded music, giving rise to the musical radio formats that still dominate the medium today.

In Radio: The Book, veteran broadcaster Steve Warren describes this period as “the great music shift.” He compares it to the “consonant shift” that spoken languages like English and German underwent during the fourth and fifth centuries, where unvoiced consonants became voiced and vice versa. His book is meant to provide practical advice for people who actually work at radio stations, not a historical overview of the medium, so he wasn’t stressing this for academic reasons. In his experience, many listeners who came of age before the shift refused to embrace “postshift music” even after decades had passed. This is how he describes what it sounded like:

• Big bands and big orchestras shifted to smaller, stand-up stage bands.

• Horns and strings gave way to electric guitars and bass.

• Wholesome gave way to a grungy, doper image (Doris Day became Janis Joplin).

• Lyrics shifted from simple to complex.

• Lyrics shifted from sentimental or romantic to harsh and realistic.

• Voices shifted from smooth and mellow to harsh and naturalistic (Bing Crosby became Bob Dylan or Rod Stewart).

• Articulation shifted from clear and accessible to difficult and sometimes obscured.

• Vocal power shifted from strong to amplifier-boosted and sound-processed.

• Love and romance became more frankly sexualized.

• Rhythms that may have been slow or moderately fast became faster.

• Emphasis on melody shifted to emphasis on rhythm.

• Lyrics became more poetic in the sense that they became more personal to the artist and, often, more sophisticated. (Moon and June gave way to contemplated and dilated.)

• Pieces as a whole became less immediately accessible but more rewarding during repeated careful listening.

Warren is describing the sound of music publishers being replaced by record companies as the dominant force in the music business. Once records, rather than songbooks, became the primary means through which the mass audience bought songs, live performers had to start focusing on recreating the exact sound of specific recordings onstage rather than interpreting a piece of sheet music in their own style.

A hit record could suddenly transform a single performer with a small electric backup band into serious competition for large dance orchestras on the touring circuit. Audiences became very interested in the specific creative decisions and personal stories of individual performers rather than the broad, accessible songwriting favored by Tin Pan Alley. Crowd-pleasing crooners like Bing Crosby and Frank Sinatra, who made a living singing everything the audience wanted to hear with no complaints, were replaced in the firmament by complicated, difficult singer/songwriters like Bob Dylan who resented feeling pressured to play anything they didn’t want to.

The entire idea of a “recording artist” itself is younger than Joe Biden. It’s a business model that was made possible by specific technology, and it didn’t really enter the mainstream public consciousness until after that technology achieved mass adoption. Even our modern conception of what a “singer” is owes much more to the invention of amplified condenser microphones in the twenties than anything people were doing with their voices before that. The notion of a special, unique talent who makes recordings of themselves for a living made the most sense in the sixties, when vinyl albums could move millions of units at a price point equivalent to sixty dollars in today’s money. Does it still make sense now that everything about the way music is made, distributed, marketed, and sold has completely changed?

Last September, Rihanna announced that she would be the featured performer at the next Super Bowl halftime show. Many of her fans were elated, assuming this meant that a new tour and album were in the offing. The anticipation only got more intense in subsequent months when she went on to announce that she would also be contributing two songs to the Black Panther: Wakanda Forever soundtrack. Rihanna had effectively been retired from music since completing a world tour for Anti in twenty sixteen, and fans hoped these new projects were an indication that her long hiatus was finally coming to an end. That hope was misplaced.

The conventional wisdom about “artist careers” in the post-streaming era is that touring has become the principal moneymaker. When Spotify introduced their all-access subscription product, the public swiftly became accustomed to spending less on recordings than they had been. The consensus that emerged, about how touring is the way artists must make money now, feels true to streaming subscribers because it accurately reflects their changing preferences.

If this were true of Rihanna, though, it would make no sense to let a such a massive promotional opportunity pass by without announcing a new tour and album to capitalize. The Super Bowl halftime show is the most-watched television event of the year. Lady Gaga’s Joanne album was basically dead before the NFL’s marketing muscle allowed Interscope to turn “Million Reasons” into a respectable radio hit. No better platform exists for artists in the U.S. market who want to engage the largest possible audience of potential customers all at once. Why not take advantage?

It’s simple. The mass audience may be willing to pay for tickets. They may even be willing to stream individual recordings an obscene number of times. Even then, the total number of dollars that fans are capable of putting on the table is insufficient to compete with the other income streams that are available to an artist of Rihanna’s stature. She is the perfect example of what the “recording artist” has become in the streaming era because recording is now utterly peripheral to her overall business. She doesn’t work for fans anymore. Our collective desire to see what she will do next is functionally just a bargaining chip in negotiations with the only audience that really matters anymore: multinational corporations like Disney, which owns both the Black Panther franchise and the broadcast network on which the Super Bowl airs.

They must have been pleased! This year’s Super Bowl telecast was the most-watched in history, and Nielsen ratings indicate that the halftime show was seen by even more people than the game itself. Rihanna’s brand is now stronger than ever, and the next cancerous multinational to decide they need a global pop star to serve as a temporary mascot for their products will have her at the top of their list. Side ventures like her apparel and cosmetics lines with higher profit margins than records and concert tickets will continue to be her primary focus. The individual creative decisions by “artists” that once drove interest in physical music formats now generate much greater dividends when applied to products that have very little to do with music at all.

In fact, corporate money was the lifeblood of many artist careers during the twenty tens. Last year, the electropop artist Santigold was forced to cancel an entire upcoming tour in order to avoid losing money on it. When discussing the situation in an interview with Variety, she noted that it was “private” and “corporate” gigs that had been making her past tours profitable. As those opportunities dried up, it became impossible to continue approaching her career the way she had been. The energy drink conglomerate Red Bull was such an aggressive, visible proponent of the UK electronic label PC Music early on that when A.G. Cook and Sophie collaborated on the debut single by QT—a mysterious project pitched as a pop star who is also a sentient energy drink—it wasn’t immediately clear that the whole enterprise wasn’t some kind of sponsored content. Ad agencies working on behalf of brands like Doritos, Mountain Dew, and Converse were showering money on indie acts back then.

Many musicians from the era of iPod commercials and SXSW showcases even opted to retire from art and chase corporate work full-time. A BuzzFeed News investigation from ten years ago (entitled “How Selling Out Saved Indie Rock”) profiled members of the Los Angeles-based indie rock band Darker My Love, who decided to pivot fully to jingle-writing after enduring a grueling run of tour dates playing as backing musicians for The Fall’s Mark E. Smith. The smooth, predictable business of crafting “soundalikes” and downtempo covers of old pop hits for movie trailers turned out to hold just as much appeal for corporate clients as it did for the musicans themselves. That kind of work-for-hire commercial audio has been the de facto sound of advertising for a decade, turning a page on the era of car companies licensing old record collector favorites like Nick Drake’s “Pink Moon.” The surest path to success in this economy, whether you’re a celebrity brand or a working musician, is to escape the traditional music business entirely. Like rats fleeing a sinking ship.

Last month, an anonymous producer known only as “Ghostwriter” showcased a new Drake song on TikTok. It’s called “Heart On My Sleeve.” Clad in a bedsheet and some clout goggles, they described the song as having been “made with AI,” but that framing is a bit misleading. We know what a Drake song created entirely from scratch by generative tools like ChatGPT would sound like, because a VC-backed startup called Mayk.it released an app that could spit one out on demand earlier this year. The results were flat, aimless, and deeply unimpressive. “Heart On My Sleeve” is different. As a piece of songwriting, it is no more or less remarkable than the thousands of other attempts to recreate Drake’s success that get submitted to streaming platforms every week. Like Cher’s “Believe,” the only genuinely unprecedented thing about the record is the way the vocals sound.

In fact, “Heart On My Sleeve” has nothing to add to any of the ongoing arguments about whether or not LLMs can replace journalists and screenwriters. In every way that matters, a human being wrote it, produced, and performed it. What the song’s viral success does tell us, if anything, is that the mainstream inflection point for Vocaloid software foretold by Bill Werde and Ed Stratton in the New York Times twenty years ago may finally have arrived. An independent creator working without direct assistance from tech companies, record labels, or artist teams was able to build and operate a DIY voicebank based on an established celebrity.

''It is a matter of time before Yamaha makes this technology available for consumers to make their own fonts,'' Mr. Stratton said. But at present, the process, which requires a deep knowledge of phonetics and audio engineering, is too complex for ordinary consumers. Even if an ingenious audiophile were to untangle the process, however, he would still need a database of thousands of articulations -- more than someone would be likely to cobble together from available recordings. As for famous voices now lost to time, if they left behind a substantial enough catalog, it might be possible to produce at least a portion of the required phoneme database. The rest of the required vocals could come from a sound-alike singer.

The exact specifics of Ghostwriter’s workflow are not currently known, but there’s not really any mystery involved. It’s Vocaloid. They assembled a library of high-quality recordings of Drake’s voice in order to synthesize a “new” performance from them. Whether or not Ghostwriter personally thinks of their Drakeface production stack as being equivalent to Yamaha’s “Vocaloid” or Crypton Future Media’s proprietary new “Piapro” platform is irrelevant. A computer did not “generate” full-on songwriting the way ChatGPT “generates” blocks of text. There is not a sentient digital facsimile of Drake living on a server blade somewhere that can spit out tracks like “Heart On My Sleeve” on command. The big innovation here is just the interface. A bunch of nerds in a musty Discord server somewhere have made Vocaloid easy enough to use to justify doing illegal things with it. It has crossed the accessibility threshold that Auto-Tune had to breach in order to become so instantly ubiquitous in the aughts.

Creating Vocaloid “performances” used to require manually drawing and editing each note and syllable on a grid. According to a Vice interview with a scammer who sold fake Frank Ocean demos to a leak server, today’s DIY voicebanks can pull all of that information out of an existing recording. You can feed it a clip of yourself singing anything you like, and the software will attempt to automatically detect and swap out your phonemes for Drake’s. It will not compose a melody for you, nor will it help you to stay on-beat or place ad-libs effectively. It will not tell you how to rhyme “lay low” with “table,” or explain why you would want to. A celebrity’s phomemes can’t make you a great songwriter or performer any more than a verified badge can make someone good at posting.

Producer and songwriter Max Martin has always justified clunky, nonsensical lyrics like the chorus of Britney Spears’ “Hit Me Baby One More Time” by arguing that the legibility of a lyric does not matter if the words sound good together. That’s the part that’s tricky to replicate, which is why Martin’s brand remains so sterling despite the infamously fickle, pernicious nature of the record business. Predictive text generators may be able to spit out legible rhyming couplets, but they cannot write hooks. Entirely different disciplines.

What “Heart On My Sleeve” did do for Ghostwriter is exactly what Vocaloid has been doing for the secondary creatives of the Hatsune Miku media empire for over fifteen years: it allowed them to use another person’s voice as a platform. In recent years, when a high-profile individual creator like Minecraft’s Markus “Notch” Persson falls out of favor with their own fans, people like to joke on social media that Hatsune Miku has replaced that person as the officially recognized inventor of the thing they love. So, whenever Notch’s name comes up in the news, people respond by posting something to the effect of “it’s so impressive that Hatsune Miku created Minecraft!” Since Miku is a crowdsourced platform rather than an individual auteur, this joke has the effect of redirecting credit for a piece of intellectual property’s success from the work’s author to the fans. The narrative being advanced by “Heart On My Sleeve” is a story about control of Drake’s voice being handed over to the open market, where anyone can use it to say anything. Does he want that? Should he? Should you?

Few artists can claim to have dominated streaming platforms to the extent that Drake has over the past decade. He figured out the recipe before anyone else: long albums, to monopolize as much of the listener’s time as possible before the algorithm sends them off to hear someone else’s record. Soothing, atmospheric production that doesn’t grate the way the bright, thrillingly annoying hooks of the radio era did. Emotionally ambiguous lyrics and melodies that feel appropriate in a variety of contexts. Melodic, digitally-processed vocals that locate a comfortable middle ground between rapping and singing. A steady cadence of releases buoyed by diaristic yet broadly relatable lyrics that read like status updates. My personal favorite song of his literally does just seem to be a first-person stream-of-consciousness monologue about looking at Instagram. Some artists arrive before the world is ready for them, others long to have been born in a different era. Not Drake. He showed up for his moment like a boomer retiree shows up for the early bird special on prime rib.

Still, recorded music is worth less in the streaming era than it was before. Drake’s Spotify numbers may be untouchable, but his net worth started to lag behind that of his peers once they all began pivoting to new careers in acting, fashion, or private investment. He struggled to achieve a moment of transformative corporate synergy like Jay-Z’s sale of Rocawear to a hedge fund, 50 Cent’s Vitamin Water windfall, or Kanye West’s transformation into a cyberpunk Dov Charney for the AliExpress decade. As fellow Calabasas residents like Kylie Jenner and Travis Scott debuted a billion dollar cosmetics line and high-profile “collaborations” with companies like McDonald’s and Fortnite developer Epic Games, Drake’s streaming empire started to look quaint by comparison. He was on his way to becoming the last recording artist.

Five years ago, he attempted to course-correct. Like Kanye West, Drake was working on a deal with Adidas to produce a branded apparel line. Like Beyoncé and Rihanna had done in the past, he was going to utilize the moment to introduce the world to his first child. Expert Drake-watchers quickly noticed that he was no longer wearing Nike sneakers exclusively, and no less an authority than Michael Jordan’s son Marcus fired off an ominous tweet suggesting that Drake would soon be “stuck in Adidas kicks forever.” His “October’s Very Own” brand had never found purchase outside his core fan base the way Fenty and Yeezy products had, but it looked like the upcoming deal might finally put him in a position to change that. Then, disaster struck.

Pusha T is nowhere near being in Drake’s league when it comes to sales and streams. Consequently, he’s always been quick to stress his accomplishments in other, more lucrative industries. Early in his career, this meant constantly referring to his past as a drug dealer, but today it also means bragging about his corporate resumé. He says he wrote the “I’m Lovin’ It” jingle for McDonald’s after they signed up his friends Justin Timberlake and Pharrell Williams to help craft a billion-dollar marketing campaign. He hit the campaign trail in support of Hillary Clinton in twenty sixteen. He served as the president of Kanye West’s label G.O.O.D. Music for seven years. To hear Pusha tell it, making money and securing the respect of behind-the-scenes power players has never been a problem for him, but his own records have never attracted the huge audiences that Drake’s have. Perhaps this is why he’s been going out of his way to decry Drake’s success as illegitimate for basically as long as it’s been happening.

Pusha’s attacks on Drake became mainstream talking points because his social proximity to Drake gave his accusations credibility. He claimed to know things about Drake that the general public did not, and was able to use this information to control the conversations about Drake that were taking place on gossip blogs and social media. When he learned about Quentin Miller, a struggling songwriter stuck in a predatory publishing deal that contributed lyrics to Drake’s If You’re Reading This, It’s Too Late mixtape, he advanced a narrative that Quentin was Drake’s “ghostwriter.” That his talent was the real engine driving Drake’s success. Using co-writers is a fairly standard practice, which Drake was quick to point out in response, but Pusha’s spin made the notion that Drake had stolen his career from a star-crossed underdog irresistible to the commentariat. The legend of Drake’s ghostwriter was now inexorably part of his story.

When Pusha learned about Drake’s plans to sign with Adidas and involve his son Adonis in the marketing campaign that would follow, he saw an opportunity to do even more damage. On a diss record entitled “The Story of Adidon,” Pusha used this information to paint a picture of Drake as a delinquent father who saw his own son as nothing more than a marketing opportunity. Few of the other celebrities who have sought control over when and how the media learns about their offspring have been accused of “hiding a child,” but Pusha rendered Drake vulnerable to that line of attack by revealing the information himself. The coup de grâce was the track’s cover art, a startling photo of Drake wearing blackface makeup that had been taken as part of an edgy photoshoot earlier in his career and then buried. Pusha presented it without explanation, prompting Drake’s fans to wonder what other skeletons might still be locked away in his closet.

Drake’s talks with Adidas fell apart in the aftermath. When asked about the dissolution of the deal directly, Pusha described himself as a “gatekeeper” of the practice of corporations like Adidas “embracing tastemakers and that whole energy.” The following year, Pusha signed a new deal with Adidas said to be worth “millions of dollars.” While he may not know how to market his own records to a mass audience, Pusha’s corporate ventures have given him a good feel for the ins and outs of brand management. His experience helping fast food chains to cultivate prestige helped him understand how to complicate that process for others. By using Drake’s narrative, likeness, and private life against him, Pusha was able to do real damage to his bottom line. Imagine what kind of chaos he could have caused with Drake’s voice.

After “Heart On My Sleeve” went viral, many technocrats suggested Drake should embrace the technology. If he allowed creators to put Vocaloid-generated Drake features on their own tracks in exchange for a percentage of royalties, it is said, the resulting flood of Drakeface Vocaloid music would generate passive income without requiring him to do any actual work himself. He could transition from being a recording artist into something more like a landlord, renting out his own voice to aspiring record producers. Though this sort of business model has no precedent in the western record business, the artist-as-platform model has been powering Hatsune Miku’s success in the Japanese market for over fifteen years. To understand the implications that a business model based around “secondary creativity” would have for Drake, we need only look to Miku for an example of what it looks like in practice.

Not surprisingly, this crowdsourced creativity has led to a sub-genre of sexualized Mikus, including brutal sadomasochistic motifs. There is, inevitably, a market for Miku porn. This happens with many characters in Japan, and it’s a source of some embarrassment to Itoh and Utsumi—but they don’t discourage it. The psychosexual schismatics are an essential part of her appeal, they know. “For a lot of male fans, it’s clear the short skirt that keeps flipping up is pretty important,” says Condry, who calls the erotica “another thread of the participation. People are having real emotional responses to whatever this object-character is.”

Officially, Hatsune Miku is a sixteen year-old girl. Her age, height, weight, and “outfits” are the only characteristics her creators believed it was necessary to canonize before surrending control of her to the open market. Older “fans” making songs and comics about how much they want to fuck her are as much the authors of her story as anyone else making use of the IP. Traditional artist careers are collaborative, too, but they usually involve teams working in the service of a specific vision—one that tends to involve some level of bodily and creative autonomy for the artist. The Vocaloid model is more like an online leaderboard in a video game, where a decentralized landscape of independent creators competes for engagement by any means necessary.

A constant flood of new, low-effort Drakeface content on streaming platforms would undermine Drake Prime’s efforts to make his own official new releases feel like events, which would in turn make it harder for him to promote concert tours and sell merchandise. This would be true even if creators only ever used Vocaloid to make records like “Heart On My Sleeve” that feel more or less on-brand for the artist being spoofed. What’s far likelier is that the musical equivalent of PopTingz tweets would quickly overwhelm such offerings streaming platforms. “Marvin’s Room” would have to compete with “Millie Bobby Brown’s Room.” The greasiest, most unhinged tabloid narratives from the dankest corners of the social internet could be canonized in an instant by a sharp enough piece of clickbait. The power to force those storylines into the mainstream used to be limited to insiders like Pusha T and Meek Mill, but the platform model could let anyone participate. There would be no meaningful difference between Vocaloid shitposting and official new releases.

Yes, Drake could profit from all of it if he were taking a percentage of the royalties, but recordings are just one part of an artist’s business. In an uncharacteristically lucid Twitter thread from a few years ago, Kanye West identified the “five main pillars” of a professional musician’s business: recording, publishing, touring, merchandise, and name/likeness rights. In order to understand if the platform model would be good for an artist like Drake, we have to look at the impact it would have on each one of these sources of income.

• Recording income comes from sales or streams of an artist’s records. The platform model means higher potential recording income, because the niji sousaku of independent creators could generate more new Drake music than Drake Prime would ever be able to make himself. Flooding the market with new “Drake” music could also dilute his presence on streaming platforms and compromise demand for his existing catalog.

• Publishing is the part of the industry that deals with songwriting. Vocaloid software can’t write songs, so the artist-as-platform model likely wouldn’t increase Drake’s publishing income. Ghostwriter, for example, used Drake’s voice on an original composition. Whoever wrote the song would be the one receiving royalties through their publisher, not Drake.

• Touring income has been driven by recordings since the fifties. Before that, audiences paid to see large dance bands that played a selection of the most popular songs in the country, with no particular preconceptions about how those songs should be played. Once records took over, audiences developed attachments to specific recordings and became willing to pay to see those versions performed live. The primary appeal of a Drake show is seeing the guy who made the records perform them. If Aubrey Graham is no longer the only person supplying the market with new Drake music, will audiences still pay premium prices to see him? Will they expect him to sing viral Vocaloid hits that he had no direct involvement in creating? When Kiss tribute bands play live shows around the country, they generate publishing income for Kiss by covering their songs. If Vocaloid-powered Drakeface music blew up commercially and Drake tribute acts started springing up to give audiences a chance to see those songs performed live, Aubrey Graham might not make a dime from any of it. The platform model would work best for artists who are unwilling or unable to tour.

• Merchandise is a core component of Hatsune Miku’s business model. The Good Smile Company, which owns thirty percent of 4chan, does a brisk business in officially licensed Miku figurines. This makes sense because Miku was designed from the ground up for this purpose. She was created to take advantage of a market for figurines, porn comics, and body pillows that was already well-established in Japan. Drake, on the other hand, was not designed from the ground up to sell toys. The merch on his official website is sorted by album, which implies that fans are primarily motivated to purchase Drake-branded apparel based on their attachment to specific recordings. It’s all heavily curated and often produced in limited batches, because scarcity drives demand. Embracing the free-for-all platform model would force Drake to upend his existing approach to merchandise completely.

• Name and likeness rights have become drastically more important since the all-access streaming subscription devalued recorded music. For Drake, this would entail initiatives like his Nike sportswear line and the marketing campaign for Sprite that helped launch his career. As the implosion of his Adidas deal amidst his feud with Pusha T indicates, losing control over a crucial part of his brand by licensing out his voice could have drastic consequences for this part of his business.

It continues to be the case that in streaming era, recording income is much lower than in the physical music era. Many big artists can scarcely be bothered to chase after it anymore, preferring to refrain from releasing new music until it can be incorporated into a larger plan to promote tours or clothing lines. The platform model would need to do an enormous amount of heavy lifting just to make recordings appealing as a revenue stream again, and that’s assuming it wouldn’t also do irreperable harm to the other pillars of a traditional recording artist’s business in the process. That’s not an assumption I’d personally be comfortable making, because “artist careers” have been driven by scarcity since the dawn of the recorded music era itself. As a society, we simply did not care so much about individual performers until after recordings became the primary way we all engage with music.

The artist model and the platform model are not just incompatible, but actively corrosive to one another. They cannot co-exist. I’m not arguing that one is better than the other, I’m saying that no living person will be able to do both effectively at the same time. The artist-as-platform model isn’t an evolution of the recording artist concept as we currently understand it, but a completely different proposition that would change the sound and character of popular music just as much as recordings themselves did during the “great music shift” of the fifties and sixties if it ever becomes dominant. For Vocaloid Drake to thrive, Drake the “recording artist” would almost certainly need to be destroyed. This is what advocates of “AI” are pushing for when they call for established artists to “open-source” their names, voices, and likenesses. They don’t understand how any of this works, and they don’t want to. They’re just here to break things. It’s all they know how to do.

Even leaving Vocaloid aside, the instituton of the recording artist was in peril long before Ghostwriter uploaded “Heart On My Sleeve” to streaming services. The peer-to-peer file-sharing ecosystem did afford listeners the chance to involve themselves directly in building whatever comes next, but we all collectively surrendered our leverage in those negotiations when we abandoned MP3 in favor of the all-access streaming subscription.

The all-access subscription is not, as it turns out, a viable long-term replacement for physical music formats. Spotify has never turned a profit, and competing services like Apple Music and Amazon Music aren’t expected to. The cost of their operation is basically just a promotional expense in the quest to sell iPhones and Prime subscriptions, and either service could easily be shifted to a less costly model if Spotify’s flagship subscription business were ever to fail. Music streaming as we currently understand it is just a tourniquet, a way for the record industry to temporarily stop the bleeding while it waits for help to arrive.

Recording artists are special because of their records. Even the Grateful Dead, a band known specifically for spontaneity and improvisation, didn’t truly become a mainstream force until bootleg recordings of their live shows started circulating on cassette in the eighties. The less valuable recordings are, the less special artists become. Last month, the backlash to Frank Ocean’s abbreviated Coachella set began on social media before the set itself had even ended. Those sympathetic to Ocean were quick to remind irate fans that “artists don’t owe you anything,” but the shifting economics of the music business make it clear that the reverse is also true. I wonder if people are getting tired of artists. It certainly does seem like artists are getting tired of increasing pressure to justify their own existence for steadily diminishing returns. Maybe none of this works anymore. Maybe it hasn’t been working for a long time.

What if artists could be replaced with a type of performer that was designed from the ground up to be responsive to the audience’s desires? What if it were possible to generate thousands of these performers from the reams of surveillance data and intellectual property rights that corporations are already actively hoarding? Performers who always show up on time and sing the songs the audience wants to hear. Performers who always look exactly the way audiences expect them to, who always smile and do elaborate choreography onstage. Performers who never age or self-destruct. Eternally youthful teenage pop idols that never say “no” unless that’s what the customer specifically asked for. Would the masses be interested in that?

The industry absolutely would. Britney Spears’ own family and management team conspired to use a medical conservatorship to keep her working against her will, privately referring to the arrangement as a “hybrid business model.” The logic was that Britney’s “well-being” could only be secured through “business activities” that furthered her career. This thirst for new, more profitable models to replace the traditional “recording artist” career is not unique to Britney’s abusers. It fueled the major labels’ uncritical embrace of half-baked tech fads like the NFT mania of twenty twenty-two, as well as their controversial dalliance with the “virtual rapper” FN Meka. The goal that motivated all of these mortifying decisions is only mysterious if you’re determined to see each of them as a one-off aberration. Taken collectively, these examples make it perfectly clear where we’re headed.

We know all the answers to the questions that “Heart On My Sleeve” raised because they’re the same questions that Vocaloid prompted twenty years ago. The major labels will not permit their most valuable intellectual property to be “open-sourced” and turned loose on the masses. Someone will recognize this, use the tech to start creating fictional pop stars instead, and one of them will catch on. Maybe it will be an individual running leaked code on their own machines, like the independent operators that a leaked Google memo suggests will soon dominate “AI” development. Maybe it will be a large media corporation deploying an actor’s voice and likeness to sell music to a “fandom” devoted to a particular character or franchise. The contracts those actors have to sign in order to get those jobs probably already contain language permitting the studios to do this.

Either way, an atomized network of aspiring songwriters and producers will embrace whatever heinous royalty splits the character’s owners demand and tout the arrangement as a liberatory, democratizing force in a stagnant industry full of gatekeeping and nepotism. Legacy artist brands will continue to have value as long as audiences continue to be attached to their recordings, and the record industry will work tirelessly to protect that value with lawsuits and DMCA claims. As legacy artists die off, whatever replaces them will probably look a lot like Hatsune Miku.

If I had to choose one person to designate as the first recording artist, I would pick Louis Armstrong. In part because of his small band recordings from the twenties with the Hot Fives and Hot Sevens, but mostly because I think Armstrong himself would have approved. Though the bulk of his career was spent re-interpreting Broadway show tunes and Tin Pan Alley songs during the publishing era, he was obsessed with recordings. After he acquired a portable tape recorder in nineteen fifty, back when such devices represented the cutting edge of audio technology, home recording became his primary hobby.

In the last two decades of his life, he made hundreds of tapes. Sometimes he would pass the recorder around at a party to record his guests, other times he would just let the tape roll while was smoking weed and listening to records by himself at home. He would even create custom artwork for each tape by collaging clippings from magazines and newspapers, as if he was planning on posting them to SoundCloud as DJ mixes. He left a huge archive behind after his passing in seventy-one, and his sincere enthusiasm for the tech radiates from it even today.

To understand why Armstrong was so passionate about recordings, we have to look at the world he came up in. When his first recordings with the Hot Fives and Sevens were released back in the twenties, they were categorized as “race records.” In the context of the present-day music industry, the idea of an officially recognized genre category just for Black music feels insulting, and conjures up images of racist record companies concocting evil schemes to minimize the accomplishments of Black artists and keep them out of the white mainstream. Given the way terms like “urban” and “latin” are constantly deployed in the present to justify the hyper-consolidated American radio landscape’s demonstrable fixation on white, English-speaking artists, this feels like it would be a reasonable assumption to make. As we’ve discussed, though, the publishing-era music industry was very different from the one we know.

In the publishing era, the pop mainstream revolved entirely around songbooks penned by professional songwriters working in New York City office buildings. Those were the songs that Armstrong was getting paid to play every night back when his main gig was playing in big orchestras led by bandleaders like King Oliver and Fletcher Henderson. He cut those early small band records as a side project in between the live dates that actually paid his bills. Records were still a niche product compared to sheet music, and “race records” were among the most popular records on the market. For fans of this music, these records provided an extremely welcome alternative to what the publishing ecosystem was offering.

What exactly was the publishing ecosystem offering, then? What could possibly be bad enough to make the concept of “race records” seem progressive by comparison? Why was Louis Armstrong, one of the biggest success stories of the publishing era, so elated to see it end?

Blackface. The answer is blackface. In the eighteen hundreds, traveling “minstrel shows” featuring white men in blackface experienced enormous success capitalizing on white Americans’ fascination with Black culture. For most of this country’s history, Black culture was represented in the mainstream primarily by white minstrels, who used it to paint a bizarre, distorted portrait of life under slavery that was supposed to make it seem like it had been idyllic and peaceful. Minstrel shows were a key influence on the development of vaudeville, and vaudeville was consequently lousy with minstrels. Before the advent of radio, Tin Pan Alley relied heavily on vaudeville performers to turn new songs into hits, which often meant penning the nostalgic ballads that minstrels used to make their noxious case to audiences that everything was better and simpler back in the good old days.

When the silent film era ended, motion picture studios bought up all of the music publishers to ensure they’d have enough music to soundtrack the “talkies” they were about to make, and the minstrel tradition followed the publishing industry to Hollywood. The first such feature to be released by a major motion picture studio was The Jazz Singer, a musical drama about a professional blackface singer. The film starred Al Jolson, a vaudeville veteran often called “the king of blackface performers” by historians. A biography of him from nineteen eighty-eight suggests that the makeup revealed “an impudent and joyous harlequin” lurking within him, and that his blackface performances demonstrated “an elan no other performer—black or white—would dare exhibit.” It would be difficult to overstate how popular this stuff was, or how difficult it would have been to counter the influence it had before recordings gave artists like Armstrong a chance to speak to the masses with their own voice.

This is why Armstrong loved to record himself and the other important people in his life. He believed recordings would allow them to be remembered as they actually were. As long as his recordings endured, he thought, ignorant whites wearing cartoonish disguises would never be able to twist his story into a narrative that reinforces the lie of white supremacy. His life and works would never become raw material for propaganda designed to undermine his very humanity. “Authenticity” is treated like more of an abstract concept in music today, but to Armstrong, it was all-important. To him, the idea that authenticity matters was the silver bullet that killed the minstrel tradition. Ever since then, Black music has been America’s most beloved and significant cultural export. It’s basically the only thing the rest of the world likes about us, and they experience it primarily through records.

Armstrong’s tapes make it clear that he was deeply concerned with the racism he encountered in the entertainment business. As such, I have to wonder how he would respond if it were explained to him that in the future, his vast archive of homemade tapes and professional recordings could be used to create a digital facsimile of his voice. That anyone who wishes to will then be able to use their own voice to control his, outputting new “recordings” of his voice saying things he would never have said in life. That some anonymous figure literally wearing a white sheet to conceal their identity has already done this two of the pre-eminent Black entertainers of our time. What would someone who experienced the liberatory potential of new technology on that level think about all of this? Would it look like progress to him, or just an elaborate way to bring minstrelsy back?