The Next Spotify
AI provides aspiring songwriters around the world with the opportunity of a lifetime. Smart, fair licensing is the key to the new music rennaissance.
“AI voice will get better, and listener resistance to it will, accordingly, shrink. The question with game-changing new technologies isn’t even if, but when.” — CNBC
Napster 2.0?
We’ve seen this movie before, right? The music industry is blindsided by new technology and threatens massive lawsuits while playing a futile game of Wac-a-mole with millions of rabid music fans downloading every song they ever wanted for free. 40 years ago the compact disc (CD) was introduced into the US marketplace. The conversion of music from analog to digital changed the music industry and created a massive cash cow for the major record labels as music fans “upgraded” their existing vinyl music collections to digital. The industry was riding high and then….Napster! The gluttonous digital music money grab by the record labels had officially come back to bite them in the ass. Well, at least digital music spared us from thousands of horrific album covers. Two decades after the peer-to-peer revolution by Napster, the music industry now faces another existential crisis, the advent of artificial intelligence (AI) as an ideation, production, mixing, mastering, distribution, and promotional tool for millions of aspiring music producers worldwide. And again, all for essentially free.
Arguably the latest “Napster moment” for the record industry occurred when @ghostwriter977 dropped “Heart on My Sleeve”, a track utilizing AI-generated voice recreations of the voices of pop music icons Drake and The Weeknd. The track caught on and generated over 20 million streams on Apple Music, Twitter, TikTok, Spotify, and YouTube in the first 48 hours after release. In response, Universal Music Group lost their proverbial shit and reached out to Spotify and Apple Music to remove the track and request that they effectively delete “Heart on My Sleeve” and ban any other AI-generated music.
But was it already too late? Is the friggin’ cat out of the bag? Novelty or not, AI-assisted songs are becoming popular on the internet, similar to the beginning of peer-to-peer sharing was on Napster. And it’s still early, we are at the top of the first inning in regard to the impact of AI and the stakes are sky-high for the entire music industry. It’s a big, controversial topic and there is no shortage of opinions about the impact the tech is going to have, with some artists critical of AI music while others say AI could usher in a new renaissance , some saying it sounds “mad as fuck” and others saying there’s nothing to worry about until AI “trashes a hotel room”.
I get the controversy, but is it that much of a reach for people to appreciate hearing iconic music voices within a new (synthetic) context? For some, yes, and I understand. It’s a deeply ethical, philosophical topic. It also will impact careers and livelihoods. Everyone’s perspectives should be respected. And yes, copyright and fraud are a huge issue right now. But these issues will eventually get sorted out. In the meantime, let’s focus on the power of voice cloning and the immense opportunity.
While the 3 major record labels are staring at a technological threat to the profits made from their cash-cow headline artists, AI voice cloning also presents them with a wild twist in the form of a potential bonanza of lucrative licensing opportunities, provided that the labels can thread the needle properly. What do I mean by threading the needle? Basically, going forward, the labels and copyright owners will need to find the right balance of a controlled release schedule of well-produced, original AI voice clone songs from select artists in their catalogs to capture the attention of the media and curious customer base to boost profits while simultaneously not allowing the market to become flooded with second rate AI cover songs and bad (or potentially offensive) songwriting that jeopardizes the brand value of their core music assets.
In a nutshell, fight fire with fire and stamp out the rush of AI cover songs. Provide the creators with a dedicated audio production platform and tools to generate high-quality AI vocal clone songs provided that they use original song lyrics that can be protected by copyright. Pre-existing songs from old bands and songwriters that already have copyright protection or brand-new lyrics written by the aspiring producer would be eligible for collaborative publishing. Collaboratively produced songs (AI vocal clone singers and original song lyrics) generated on the platform would be used as “demos” to submit to the estates or rights holders of the singers they used in their voice-clone songs (this is already a growing trend among professional songwriters). In other words, produce and release better copyright-protected and licensed AI vocal clone songs coupled with the opportunity to potentially make some real money. Similar in ways to what the record labels did against Napster when they struck partnerships with Apple, Spotify, and other platforms that offered the same online convenience for a small subscription fee (probably too small) without the risk or hassle of prosecution from downloading “pirated music”.
“Starting around 2000, US music revenue fell off a cliff — from a peak of $21 billion in 1999 (in 2018 dollars) to about $7 billion in 2014, according to data from the Recording Industry Association of America.” — Quartz
While GhostWriter was clever to pick two of the most streamed artists on social media and thus attain massive attention to quickly scale to millions of streams and views, what was also important was that the track used original lyrics and was not just Rihanna covering a song by Beyonce. I would argue that its streaming success was at least partially due to the original lyrics which led people to believe that a brand new track had been dropped by Drake and the Weeknd. It was also well produced with pretty good sound quality, which greatly helped its credibility. While some people may have been duped into thinking it was an original track sung by Drake and the Weeknd, many others may have suspected it was AI-generated “FakeDrake” but still dug it anyway.
This is a critical point of my argument in this post. That music is subjective, and while some people will shun AI-generated music, others will embrace it as part of the 21st-century music ecosystem. Not to generalize but it would seem natural that young music fans would be more receptive to experimentation and a recent study by addictive tips lends support to the theory that younger audiences are generally more accepting of voice experimentation/AI versions of their favorite artists.
Right this moment, more AI voice cover songs are being easily created and uploaded to a variety of streaming platforms. Millions of daily views, almost zero regulation, wild west industry chaos, and nobody is making squat. Here are some examples of voice cloning tech used to create major (deceased) artists covering classic songs from major contemporary artists (some also deceased).
Frank Sinatra
Creep (Radiohead), Livin’ on a Prayer (Bon Jovi), Bohemian Rhapsody (Queen), Gangsta’s Paradise (Coolio), Careless Whisper (George Michael)
Whitney Houston
Bohemian Rhapsody (Queen), Someone Like You (Adele), I Just Can’t Stop Loving You (Michael Jackson), Listen (Beyonce)
2PAC
P.I.M.P. (50 Cent), The Message (Nas), Isis (Joyner Lucas), Your Life (Kasinova), Smokin’ (Nas),
Elvis Presley
Baby Got Back (Sir Mix-a-lot), I’m on Fire (Bruce Springsteen), My Heart Will Go On (Celine Dion), Perfect (Ed Sheeran) Billie Jean (Michael Jackson)
Like it or not, just for the novelty aspect alone, there will always be a black market for AI clone songs going forward. What must be avoided is the release of a smash hit by an A-List artist and then a hundred different voice clone versions flooding online the next day. The RIAA and legal forces they unleash will attempt to shut down deep fake “fraudsters” and perhaps slow the onslaught via legal recourse, but that will just be followed by a new wave of underground AI tracks being released to social media, the dark web, streaming services, and across the wild, wild web, creating a vicious, never-ending cycle. But on the other hand, let’s be real. The major record labels are much more concerned about voice cloning of their big cash cows such as Drake, The Weeknd, Rihanna, Taylor Swift, etc than they are about their lesser-known, longtail assets. This is understandable and in itself presents the opportunity for experimentation with various record label rosters of one-hit wonders that are well on their way to fading into obscurity. The trillion-dollar question is, can the best AI voice clone songs be finely curated and monetized by the industry?
I believe the mechanism required to adapt to a flood of rampant deep fakes and seize the emerging opportunity is an advanced online music platform marketplace (the Next Spotify) that allows for the full participation of aspiring songwriters and producers and provides them with a legitimate path to attain both social accolades and financial rewards significant enough to actually make a living. For millions of starving songwriters, AI voice generators could be the stepping stone they need to break into the industry. Over time, this may include utilizing existing tracks from previous albums that have been collecting dust, in some cases for decades, and combining them with iconic voices of legendary singers. Here’s how…
For starters, from an archival perspective, existing song lyrics from an older indie band with copyrighted songs may be used. AI might be developed that has the ability to search much more accurately and curate song lyrics based on theme, mood, feelings, and things beyond genre and category (pop, rap, metal, country, etc). This would allow aspiring producers to more accurately match a cloned voice with an existing copyright-protected song that fits their vision. The platform would thus introduce new revenue for participating copyright owners and aspiring record producers. Tens of thousands of dormant indie bands and performers of the past would have a new opportunity to make money off their previous songwriting and compositions via collaborations with innovative producers that combine their original song compositions with the voice clone talent of emerging or established singers. AI voice cloning collaborations provide an opportunity to repurpose their existing content and (if a collaboration song were to catch on with mainstream artists and go viral) potentially make a tidy sum and rejuvenate their careers. Bottom line, people can debate the ethics of voice cloning a deceased artist ad nauseam, but once a brand new, well made, bumpin’ (AI) jam from Tupac Shakur drops, streaming numbers are going to spike, big time.
“….there will be people 20 years from now, [saying], ‘Oh, I much prefer AI Rolling Stones than [the original] Rolling Stones. That’s just gonna be a thing.” — Rick Beato
Who are these aspiring producers and how many are there? Let’s just conservatively guesstimate several million young (and some old), tech-savvy hobbyists, superfans, and aspiring producers around the world who are ready to pounce on a once-in-a-lifetime opportunity presented by evolving AI production tools combined with voice cloning technology. And while these tools may have been only previously available to studio engineers, they have recently exploded online and into the global collective consciousness of music enthusiasts around the world following a number of shockingly good AI songs going viral on social media. Artificial intelligence and voice cloning technology represent the potential for widescale democratization of music production and an industry renaissance via the creation of an expansive new genre category, the AI vocal clone song.
“We are OK with autotune to fix vocal inconsistencies and make a singer’s voice sound better. We are OK with lip synching in concerts. We are OK with some sampling and with unnamed musicians playing instruments. We are OK with ghostwriters and hit-making producers. So why are we drawing the line at AI-created vocals?” — The Hill
Did I mention “localizing content” or the ability to translate any voice into other languages? Yep, they got software for that too. Now, billions of music fans will be able to listen to The Beatles, Ozzy Osbourne or Little Dragon in their own language. Wrap your spinning head around that for a minute. While voice translation has existed for basic educational translation services, the tech has recently drastically advanced to the point where it can be used in music videos. The AI cover song possibilities are endless when you see the forest through the trees. With literally millions of independently produced songs dating back to the seventies (and beyond), the opportunity to create a vibrant AI voice clone song marketplace with the collaboration of independent record labels alone is a viable opportunity.
But what is really mindblowing is the combination of original songwriting from bedroom-based producers combined with an unlimited number copyright protected AI voice clones that is possible within a dedicated, subscription-based online platform. Far fetched? AI voice clone platform models that incorporate copyright protection are already starting to bubble up.
In this post, I’m gonna attempt to connect lots of dots and demonstrate the viability of what I believe is a billion-dollar AI music aggregation, production and distribution marketplace enterprise, the Next Spotify.
Ready?
Down the rabbit hole we go….
$ubjectivity
Check out this list of names. Any favorites?
Tina Turner, Sharon Jones, James Brown, Maurice White, Etta James, Amy Winehouse, Marvin Gaye, Ray Charles, Fela Kuti, Bob Marley, Ronnie James Dio, Lemmy, Bobby “Blue” Bland, Martha Reeves, Wanda Jackson, Sam Cooke, Whitney Houston, Harry Nilsson, Johnny Cash, Merle Haggard, Loretta Lynn, Buddy Holly, Tammy Wynette, Elvis, Patsy Cline, Teddy Pendergrass, Barry White, Billie Holiday, Nina Simone, Frank Sinatra, Tony Bennett, Astrud Gilberto, Karen Carpenter, George Michael, U-Roy, Peter Tosh, Michael Hutchence, Aaliyah, Curtis Mayfield, 2PAC, Biggie Smalls, Eazy-E, Heavy D, Nipsey Hussle, Guru, Chet Baker, Freddie Mercury, Lou Reed, David Bowie, Louie Armstrong, Billy Paul, Luther Vandross, Gladys Knight, Selena, Chris Cornell, Sinead O’Connor, Janis Joplin, Wilson Pickett, Roy Orbison, Jimi Hendrix, John Lennon and Prince.
Who is better, Tina Turner or Etta James? Johnny Cash or Tony Bennett? Barry White or Curtis Mayfield? Those are subjective questions.
All of these incredible musicians and singers have passed away and I (and hundreds of millions of others) still listen to and enjoy their music weekly if not daily. Their songs are classics and their voices are timeless and dare I say, immortal. They represent the power of nostalgia and business knows that nostalgia sells. In some ways, listening to the classic rock, new wave, and old-school jams is the closest thing I have to a Hot Tub Time Machine to hold on to my past and I’m sure millions of others feel the same way,
How many millions of others? That’s another subjective question.
Speaking of subjectivity…
Music is subjective. You might call a symphony orchestra performance at Carnegie Hall good, “proper” music while I might be just as enthused about the percussion from a guy playing drums on a bunch of empty paint buckets on a busy corner in SOHO. That’s subjectivity, the quality of being based on or influenced by personal feelings, tastes, or opinions. Some people may say that music created using artificial intelligence has no emotion or human expression, while others may embrace AI-generated music voice clones because they just love an artist’s voice and don’t care if it’s not the actual person singing. To each, their own. Ask a thousand people what music means to them personally and get a thousand different answers. Subjectivity is a critical aspect of the opportunity for record labels and rights holders because acceptance of AI music by fans translates directly to streaming subscription dollars. Bottom line: Some people won’t like AI music that clones the voices of iconic singers while some will love it. Note: “some” = tens of millions of music fans worldwide.
But perhaps even more important for the majors (and streaming platforms) is how AI vocal clone songs will (re)introduce classic voices in new ways to new audiences worldwide. Younger demographics that are impossible to reach because they are hopelessly caught up in the throes of the attention economy, pre-occupied with dozens of activities including gaming, movies, sports, gambling, booze, weed, influencers, fashion, television, VR, social media, investing, (not to mention) dating, mating, babies, holding down two jobs, night school, side hustles, taking care of aging parents, freaking out about the climate, you get the picture. However, once exposed to a legendary singer or band via an AI voice clone song, they will find time to stream more (existing) tracks if the artist they heard tickles their fancy. This is not to say large swaths of younger audiences won’t also embrace AI vocal clone songs from their favorite contemporary bands and singers, many definitely will. But the point is that AI extends past artists’ shelf life immeasurably and record labels and stakeholders should be elated.
NEWSFLASH: This is called repurposing your existing content, and it’s marketing 101. One might also argue for the music industry, it’s a quasi-new form of covering your ass (CYA) similar to the film industry constantly releasing sequels and rehashing successful franchises in order to capitalize on “built-in audiences” already familiar with the subject matter, thus eliminating the need to burn millions in marketing dollars to build awareness for new concepts and reducing their risk of releasing a flop.
Taking it a step further, this is also known as brand extension and yes, it most definitely crowds the existing market and puts even more pressure on new, aspiring artists to get noticed and build an audience. However, the Next Spotify may also offer a new type of opportunity for newbies to break into the industry as an AI voice clone songwriter and/or producer.
QUESTION: How many angst-ridden sixteen-year-olds with spiral notebooks full of deeply poignant, personal thoughts, poetry and incredible song lyrics are there on earth? No longer will these legions of ambitious yet self-conscious creatives be held back because they can’t sing like Beyonce, Taylor Swift, or Adele. All these digital tools and software are now at their fingertips and voice cloning combined with writing original song lyrics might unlock a tidal wave of artistic self-expression and provide a potential fast track for undiscovered talent to gain online fame and celebrity.
They might even get started by just singing to their Alexa.
“Songwriters have been getting the short end of the stick in the music industry for a long time. And the reason for that, for many of them, is that they’re not able to produce. So they write down lyrics [and the bones of a song] using a guitar or a piano, then they send it to the producer or the artist. Then that producer creates a new track, and then the artist releases it. And the songwriter only shares on the publishing, not on the master.” — Diaa El All, Soundful
The Three Signposts
There is now, no question that artificial intelligence (AI) is a game-changing technology. It is already disrupting a wide range of business verticals including the multi-billion dollar music industry. From potential copyright lawsuits by the large record labels against big tech companies for infringing their artist’s rights during the training process using Large Language Models (LLM), to software that prevents unauthorized website cloning for training, to the release of hundreds of AI music apps and tools to more adoption of the technology by record producers and as mentioned, the tidal wave of “deepfake” songs generated using AI to “clone” the voices of mainstream pop artists like Drake, Kanye, and Rihanna. The impact of AI on music is accelerating at light speed.
There are a number of ways that artificial intelligence can be used to create music. Here is a general overview with links to lists of AI music tool vendors and software.
By using prompts to create AI text-generated song lyrics (ChatGPT, BARD, and others)
Music generation and audio mastering (Amper Music, Aiva, Soundful, Ecrett Music, Soundraw, Boomy and others)
Mix and Master Tracks (LANDR, Unchained Music, Maastr, and others)
Synthetic voice creation or voice cloning (Respeecher, Resemble, Play.ht, Voice.ai, and others).
This blog post will focus primarily on voice cloning technology and specifically the intersection between human songwriting and voice cloning tech as a tool for aspiring music producers and the profound impact it will have on the music industry. My thoughts are that when you step back and look at the release of “Heart on My Sleeve” as a sign of what is possible using voice cloning, you can’t argue with the insane buzz it generated internationally and how quickly it spread on social media and the web. Many fans thought is was some of Drake’s best work in years. At the end of the day, the numbers don’t lie. “Heart on My Sleeve” was a hit, period.
Further, I believe this event was the first of three definitive signposts showing that AI voice clone music will become a massive category in the global music industry over the next several years. “Heart on My Sleeve” represents industry and mainstream public awareness of the power of voice cloning technology in music production. The other two signs are the partnership between musical artist Grimes with CreateSafe and Tunecore, representing the foundation for a functional, AI music voice cloning platform and creative hub in the marketplace (the genesis of the Next Spotify…perhaps). Lastly, the planned release of the AI-assisted Beatles song, “Now and Then” with AI-restored voice and lyrics of deceased legend John Lennon will represent mainstream acceptance (or not) of the use of AI voice clone songs. All three are critical steps in the evolution of AI music and voice cloning tech being able to “cross the chasm” to mainstream market acceptance and substantial revenue generation for labels and aspiring producers.
The argument presented is that the music streaming service Spotify is ripe for disruption by a centralized, AI-focused music platform that can simultaneously support AI-assisted music production and provide the automated legal services necessary to protect original copyright owners. The SaaS-based service will also offer online production tools and marketing resources to the growing armies of hungry independent music artists seeking to harness artificial intelligence for the development of their own musical creations and personas.
Lastly, it should be mentioned that this unique opportunity would not be possible without the relatively recent cultural phenomenon of transitioning from a passive music listening experience to active participation, ignited by TikTok.
“The music industry is on the cusp of another major structural change given the persistent under-monetisation of music content, outdated streaming royalty payout structures and the deployment of Generative AI,” — Goldman Sachs
Signpost 2: All Eyes on Grimes
Concepts for AI vocal clone song platforms are already beginning to form in the marketplace.
Last April, futuristic singer and performer, Grimes threw out an offer of 50% royalty split to producers that found creative ways to use her AI-generated vocals with an original track created by aspiring songwriters. Grimes also introduced Elf.Tech, a new AI voice software aimed at assisting musicians in replicating her distinctive voice for music creation. The platform allows users to submit their voice recordings, which are then transformed into Grimes-like vocals through the use of automated technology. These altered vocals can then be combined with electronically generated soundscapes and beats to produce original tracks that mirror Grimes’ signature style.
The Elf.Tech model has been trained exclusively on vocals and content owned by Grimes, with no claim of ownership on the submitted sound recording or composition. She envisions AI voice replication as growing in importance for the future of music, enabling artists to embrace algorithms to modify their voices and profit from the resulting creations.
“I’ll split 50% royalties on any successful AI generated song that uses my voice. Same deal as I would with any artist i collab with.” — Grimes
Enter Createsafe and Tunecore: Grimes then announced a partnership between Elf.Tech, AI development studio Createsafe and TuneCore, a self-releasing artists development partner owned by Believe, a digital music company. Through this program, artists and fans can use Elf.Tech to transform their vocals into a ‘GrimesAI voiceprint’ and include it in their original compositions. Once Grimes approves the collaboration, the music can be uploaded to TuneCore with Grimes as a featured artist, with the aforementioned 50% royalty split between Grimes and the creators. TuneCore has agreed to review the content, validate royalty splits, and distribute the music worldwide.
The partnership allows TuneCore artists to distribute collaborations created through Grimes’ Elf.Tech AI on major streaming platforms. The partnership aligns with Believe Group’s AI strategy, emphasizing consent, control, value addition, sharing, and proper identification of AI-assisted tracks. The collaboration with CreateSafe and Grimes offers an innovative and responsible way for artists to engage with AI technology in music creation.
Is this the genesis for the Next Spotify?
“Elf.tech is a prototype white label of CreateSafe’s Triniti platform, described as an AI-driven platform that the company claims “revolutionizes the way music is managed, distributed and marketed”. — Music Business Worldwide
“By joining forces with Grimes and CreateSafe, TuneCore becomes a first mover in the space, providing artists the opportunity to engage with AI technology in an innovative, streamlined process that provides tangible value and enables consent, control and revenue splits at scale. Going forward we’re eager to explore similarly innovative initiatives on behalf of our artists.” — Andreea Gleeson, CEO TuneCore
It should be noted that other artists like Holly Herndon have also embraced voice cloning and AI tools, with her own artificial voice, Holly Plus, introduced in 2021. Only members of Herndon’s DAO can profit from her voice model.
Other emerging AI voice clone music platforms and concepts
Voice cloning and licensing platform, Kits.AI reaches 1 million voice transformations milestone.
“The platform, called Kits.AI, says it offers tools for artists to safely create, license and share their own voice models. Kits.AI seeks artists’ permission when using voices for music creation, ensuring ethical use of AI voice models.” — Mandy Dalugdug, Music Business Worldwide
AI Resemble raises $8 Million from a mix of venture capital groups for voice cloning and AI detection technology.
“The tranche, which brings the startup’s total raised to $12 million, will be put toward further developing Resemble’s enterprise products and doubling the size of its team to more than 40 people by the end of the year, co-founder and CEO Zohaib Ahmed says.” — Techcrunch
Regal Investments announces $20 Million investment in WAVs.ai
“I firmly believe that traditional music companies will be dethroned in the coming years by innovative AI-driven platforms.” — Roble Regal, CEO Regal Investments
DJ Fresh launches Swap.ai using a proactive approach to copyright protection.
“Voice-Swap.ai Ltd combines the power of artificial intelligence with the human voice to offer unprecedented possibilities in song creation. Unlike other platforms that have faced controversies surrounding copyright infringement, Voice-Swap.ai stands out by prioritising artists’ rights and fair compensation for their vocal contributions.” — Whisky Kicks, UKF
“Voice-Swap sees the voice as the “new real estate of IP,” as Pelczynski puts it — just another form of ownership that can allow a participating artist to make passive income.” — Billboard
Singapore based Bandlab secures $65 Million in funding.
“The company says that over 40 million creators currently use BandLab. The app’s tools for creating music include a Mix Editor, royalty-free sounds, Mastering and SongStarter, or royalty-free compositions. Co-founder and CEO Meng Ru Kuok told TechCrunch that the new funding will be used on hiring, developing new features and “prioritizing new ways to benefit from the creator economy for artists and rights holders.” — TechCrunch
Warner’s ADA unit to distribute music from generative AI platform Boomy
“Select Boomy artists will be distributed and marketed by ADA across platforms such as Apple Music, Deezer, Meta/Instagram, Pandora, Snap, Spotify, TIDAL, TikTok and others.”
Pandora may be in the process of consolidating their brands SiriusXM and Pandora to create a music SuperApp to compete with Spotify. More to come on this…
Beat production software and marketplace BeatStars partners with AI start-up Lemonaide. BeatStars has a client base of ten million creators, producers and recording artists around the world.
“By incorporating ethical AI practices from the outset, we can revolutionize the way artists work with AI and create a positive impact in the music industry.” — Sean Gorman, COO, BeatStars.
The (first) elephant in the room of course is TikTok. who just announced their new streaming music service TikTok Music, now available in (test markets) Brazil and Indonesia (also testing in Australia, Mexico and Singapore). The clock is Tik-Toking as to when the service arrives in the US….(eh, sorry). While they have not specifically addressed voice cloning, they are really the pioneers of participatory music collaboration between fans uploading their own lipsyncing and dancing to popular classic copyrighted songs and some more obscure.
“TikTok Music will make it easy for [users] to save, download and share their favorite viral tracks from TikTok,” Ole Obermann, global head of music business development for TikTok, said during the Indonesia and Brazil launch”. — CNBC
UPDATE: (8/9/2023) The second elephant in the room is now Google, who entered into discussions with Universal Music Group (and potentially Warner Bros Music) to discuss, what else, AI music and AI voice cloning/deepfake tech and how it could generate new revenue for the record labels. Barron’s says the partnership “could transform the world of media.”
“John Legend and Charli XCX agreed to have their voices cloned by YouTube. It’s the start of an AI war among musicians.” — Business Insider
Why are these partnerships and new business venture concepts important? Because one (or perhaps a combination of several after a roll-up) may become the first platform scaffolding/foundation for a fully regulated production, licensing, and distribution platform for the creation of collaborative AI vocal clone songs, AKA — the Next Spotify.
From the Horse’s Mouth
To better understand the AI voice clone song opportunity, let’s get a quick summary from the horse’s mouth (ChatGPT). What does AI think about AI voice clone songs?
My Prompt:
Write about the financial and social benefits of being able to recreate the voices of famous singers using voice generators
ChatGPT:
The ability to recreate the voices of famous singers using voice generators holds both financial and social benefits that contribute to the music industry and society as a whole.
Financially, the ability to recreate the voices of famous singers can open up new revenue streams for the music industry. By employing voice generators, record labels and music producers can create new songs or remixes featuring iconic voices that have resonated with audiences over the years. This can lead to increased album sales, streaming numbers, and concert ticket sales. Additionally, it can extend the commercial lifespan of beloved artists by reintroducing their voices to newer generations.
Furthermore, voice generators can facilitate the production of tribute albums or live performances dedicated to legendary singers who may no longer be able to perform due to various reasons. This creates opportunities for collaboration between emerging and established artists, bringing fresh interpretations to classic songs and creating nostalgic experiences for fans. These projects can generate revenue while honoring the artistic legacy of renowned singers.
On the social front, recreating the voices of famous singers allows for the preservation and celebration of musical heritage. It enables younger generations to experience the voices of iconic artists who may have passed away or retired. By recreating these voices, future generations can connect with the timeless artistry and cultural impact of influential singers from the past.
Moreover, voice generators can also serve as educational tools. They can be utilized in music schools and workshops to demonstrate vocal techniques and styles employed by legendary singers. Aspiring vocalists can learn from these recreations, gaining insights into the nuances and intricacies that made these voices so captivating.
Beyond the music industry, recreating the voices of famous singers using voice generators has implications for entertainment and media. It allows for the creation of realistic soundtracks for films, documentaries, and video games that require the presence of specific historical or fictional singers. This enhances the immersive experience for audiences, making the storytelling more engaging and authentic.
In summary, the financial and social benefits of being able to recreate the voices of famous singers through voice generators are evident. From a financial standpoint, it opens up revenue opportunities for the music industry through new releases and collaborative projects. Socially, it preserves and celebrates musical heritage while providing educational value and enhancing entertainment experiences across various mediums.
“Instead of trying to prevent voice cloning, a better approach could be to make it available to everyone. This could be achieved by creating a safe and efficient platform that offers easy access to AI-powered music generation and voice cloning for a fee, such as a subscription. By doing so, a new business model can be established that can absorb the risks associated with AI-Generated Music & Voice cloning while allowing major companies and artists to be fairly compensated for their work” — Honore Piffaut AI-GENERATED MUSIC — Navigating Disruption : How Generative AI could affect the sustainability of major players in the music industry?
The Next Spotify
Ultimately, what we are talking about is the ability to bring together a unique combination of production software, automated legal services, online marketing, promotion, and distribution tools combined with the best recorded musical talent throughout history to create thousands of mindblowing original songs, remixes, and mash-ups. The online service would also provide subscribing producers access to instrumentals, acapella’s, outtakes, demos, and unreleased songs of participating bands and singers. Also available would be copyright-protected song lyrics from thousands of participating indie singers and bands who have album tracks available for collaborative projects. All newly created AI tracks made on the platform would have high-quality production, be vetted for full consent by all stakeholders involved, be legally protected by signed, binding contracts, utilize innovative revenue-sharing API technology, and be made available for immediate worldwide distribution.
The major record labels and copyright owners will resist at first, seeking to protect their assets from exploitation by non-professional musicians and aspiring composers. Therefore, it will take some time for the migration of cautious artists and record labels to wrap their heads around the opportunity. Most likely, smaller and mid-size labels will begin testing the AI song waters with some of their lesser-known artists, leading to a database of thousands of songs created by independent bands and artists with a smattering of artist tracks from large and mid-size record labels.
As new copyright-protected AI songs from indies become hits and get traction, consumer bias towards AI music will wane and open the door for the release of AI vocal clone songs from more mainstream artists. As it becomes clear that AI music represents a gargantuan alternative market for existing assets and little risk or expense, major record labels and their biggest pop, R&B, rap, indie rock, and metal stars will have little choice but to get involved. The market for vocal clone songs will snowball and grow as AI technology evolves and forward-thinking labels begin making their top assets available for AI cloning collaboration.
Bottom Line: As soon as the major labels figure out how to best produce and distribute AI-cloned vocal songs of their legacy/fading artists to squeeze out a few more bucks they will do it. After all, how many more U2 Box Sets do we need? How many more Dodge Caravan commercials will require the emotive trigger of classic guitar riffs and basslines of some washed-up, has-been band (in the eyes of increasingly diverse, young, target demographics)? Sad but true, boomers and genX are beginning to die off or at a minimum, focus more on their grandchildren than Elton John.
To combat the slippery slope to cultural irrelevance, copyright owners may be able to leverage the nostalgia of hearing the soothing voice of favorites like Selena, Cat Stevens or Barry White on brand-new, well-produced AI tracks while simultaneously introducing great artists to millennials and GenZ with modern lyrics and messages that actually resonate with younger audiences. Will there be pushback by irate music purists? Yah, u betcha. But money talks and there will be more than enough fans buying records and streaming AI vocal clone songs to make such endeavors quite profitable. The keys to success will be original, heartfelt songwriting that resonates with long-time fans and new listeners, quality musicianship, mastering and vocal clone song production coupled with an acute sensitivity to the perceptions of AI by the worldwide “built-in” customer base.
“Will we see artists ‘franchising’ their identity to those keen to replicate it, and copyright owners offering API-like facilities to enable AI-powered creators to leverage their works to create new material? In such a context, AI tools could be seen as the next evolutionary step in music sampling technology, whereby the original material is not merely reproduced, but reimagined.” — Nick Eziefula, Partner at Simpkins LLP
The way forward must be a new democratized music ecosystem that is inclusive to aspiring artists, and provide real opportunities (somewhat similar to TikTok) but in a more controlled, e-commerce focused methodology and framework. A platform designed with options to encourage, protect and preserve human songwriting, that also allows the indie labels and majors to identify diamonds in the rough from aspiring producers, automatically assigns copyright protection to songwriters, and in turn, dissuades the legions of fly-by-night opportunists and hucksters from destroying the music industry with a tidal wave of bad content uploaded to the streaming platforms and social media outlets (Spotify, TikTok, Deezer, Twitter, etc).
“Instead of punishing copycats, labels might gain more from embracing the new tech and finding ways to monetize it.” — Kait Cuniff, The Future Party
Based on the success of “Heart on My Sleeve” and the traction we are starting to see for voice-cloned songs on TikTok and YouTube, I’m willing to bet that if a new album was produced with professional musicians, lyrics written by great songwriters and marketed with sensitivity and appreciation for each singer’s musical legacy, you would see massive media publicity and a large spike in streaming and traffic to record stores during the drop (release). Look no further than the hype surrounding the upcoming release of “Now and Then” with AI voice clone restoration of John Lennon as a prime example. IMO portions of music fans will enjoy AI songs from both deceased and current artists and begin streaming more (they already are) and the momentum will begin to snowball once better produced and a wider selection of AI songs becomes available. This will culminate in a period of explosive growth once original songwriting is combined with well produced vocal cloning (similar to “Heart on My Sleeve”) to complete global mainstream acceptance of the category. Reimagining music legends with consent and shared copyright will become a massive and very lucrative category soon.
Now let’s take a closer look at the Next Spotify platform.
“…advancements in voice cloning technology are making it possible to generate high-quality synthetic voices that can rival the most beautiful human voices (Aylett et al., 2019). This means that anyone can now create music with beautiful ai-generated voice as pleasant as those of Amy Winehouse, John Lennon, Nina Simone, or any other famous artist. This could lead to a scenario where singers are facing a much strong competition, being confronted to highly pleasant ai-generated voices.” — Honore Piffaut AI-GENERATED MUSIC — Navigating Disruption : How Generative AI could affect the sustainability of major players in the music industry
The Next Spotify Stack
The “Next Spotify” platform requires reimagining traditional music production, licensing, distribution, and promotion. It will most likely be a combination of disparate professional services to build a technology (stack) of comprehensive digital tools and legal services.
Services on the platform would include:
An AI voice generator with voice cloning capabilities. Examples in the marketplace include PlayHT, Respeecher, Uberduck and many others. Available voices for licensing would include only artists and copyright holders who agree to the rules and legal parameters of MixMashter. On the flip side, only subscribing members/producers would have access to licensable voice clones and music samples to use in the creation of their demos. A wave of even more innovative voice cloning technology is on the horizon and future versions of voice generators will be relentlessly tweaked for improved sound and be capable of perfectly capturing the full range of any singer’s voice.
An online cloud-based Digital Audio Workstation (DAW) that allows you to drag in your selected tracks, samples, and resources to a digital studio environment to lay down the foundation of your next musical creation using in-platform music production tools. Producers can learn, tinker and then create full tracks using the virtual lab. Examples include BandLab , Amped Studio , Soundtrap by Spotify and others.
The Music Vault. A digital library (database) of licensed song lyrics, musical compositions, beats, acapella’s, and samples from hundreds of thousands of tracks from big and small record labels around the world. All participating record labels and copyright holder materials would be available for co-licensing opportunities on the platform by millions of subscribing producers worldwide.
Important to note is that the resolution of thorny copyright issues from the innovation of the AI revolution will hopefully open up wide-scale sampling from mutually interested parties as well. This aspect of the Next Spotify platform should not be underappreciated. It goes without saying that a major component of the genius of early hip hop music was completely stifled by the short-sighted, suffocating copyright laws by a less-than-visionary legal system that was/is far outdated for the times. In regards to sampling, let pursuing the magnificence of 3 Feet High and Rising and Paul’s Boutique be the end goal here.
Copywriting, publishing, licensing. The most important component of the Next Spotify is the automated legal services that cut out pricey attorneys and streamline the shit out of the contract negotiation process. This must be baked into the architecture and design from the get go. The platform will provide full legal copyright protection of your new song through the use of automated attribution software that assigns percentages of ownership and subsequent royalty revenue to each contributing party once the asset/track is vetted and approved by all stakeholders. While some licensing deals will require complex frameworks including concessions, rights and permissions, etc, the majority will be straightforward blanket licenses with standard boilerplate terms and conditions. Some examples of online services focused on rights and permissions include EasySong , SongTrust, and an interesting single license service model called One License among others.
What is copyright? Here is a good definition from DataIntelo:
“Copyright is a form of protection granted by the laws of the United States to authors of original works of authorship, including literary, musical, and dramatic works. Copyright protection subsists from the time the work is created in fixed form. The copyright owner has exclusive rights to reproduce, distribute copies of, prepare derivative works based on or perform publicly the copyrighted work during his or her lifetime and for a period of 50 years after death.”
Guidance from the U.S. Copyright Office
Personalized Dashboard. A handy, customizable dashboard that offers numerous modules. See “Recording a Demo?….”
One might argue that, why can’t the existing Spotify just become the Next Spotify by adding a specific category for AI songs? The answers are that
- The critical piece of automated legal services is a dealbreaker. It must be baked into the platform from the ground up and would be very difficult for any platform to try and “bolt on” as an added feature.
- There must be quality control of the content that is allowed on the platform. The world does not need a thousand poorly made Drake songs produced on a whim using voice cloning technology. It’s bad for Drake, it’s bad for the industry and most importantly, it’s bad for music fans. Whether or not it’s officially a violation of copyright violation, it remains a violation of artists and that is wrong.
“Given the legal challenges in proving that AI created music from copyrighted works and the technical difficulty of reverse-engineering a neural network, a blanket license would more simply allow artists to claim a portion of the benefits. Artists could register their music with a collective rights organization like Broadcast Music, Inc. The entity could then license their entire music catalog to AI companies for any variety of technological and innovative purposes.” — Sona Sulakian, American Bar Association 2022 ELI Writing Competition Winning Essay, Protecting the Artist: Licensing in an AI-Generated Music Market
Special Features and Modules
Imagine opening a music app and being presented with the following:
Curated lists of the weekly top twenty AI songs in hip-hop, R&B, reggae, EDM, indie rock, country, metal, and international. You notice the new Beyonce and Aretha Franklin duet/remake of “Respect” has overtaken the red-hot Beach House — Cocteau Twins mash-up that topped the AI Top 100 chart for a whopping three days, seven hours and seventeen minutes, a new record.
Daily news and videos about the latest new AI songs, mash-ups and more from your thirty favorite music artists and producers over the past month. The Headline is a shocker, “Daft Punk enters an agreement to reimagine a female Rat Pack concept album using vocal clone voices of Grace Jones, Dinah Washington, Cher, and Madonna. The first track, ‘Read My Lips’ drops exclusively on Prime Music next Saturday!”
Streaming analytics and revenue stats for the fourteen AI-assisted tracks that YOU PRODUCED. Original songs, covers, mash-ups, and remixes comprised of lyrics you wrote, licensed AI-clone vocals from popular singers, and compositions with beats and samples you either licensed or created, all modified, and mastered using a digital audio workstation (DAW). You get a big smile when you see that your new AI song “The Bullet” co-licensed with a 90’s indie band Greazy Meal with lyrics sung by Billie Holiday is gaining traction on Prime, YouTube and TikTok.
Your online music community and marketplace. Set up a profile, promote and sell your new songs, sell merch, and get involved in the promotion of concert tours of your partner artists. Examples of online music communities include Bandcamp , TikTok , Spotify and many others.
Once your AI-assisted song/demo is ready, the ability to use a submission module that places your new creation into a “holding pen” that allows copyright owners to review your song for quality control and then reject or approve it within 72 hours. Upon approval, your demo may be re-engineered by the copyright owner (artist, label, or estate) and then packaged and distributed into the streaming marketplace, pending final revenue share negotiations and signed contracts. NOTE: This aspect of the concept is gaining traction already as songwriters are beginning to use AI vocal cloning to submit song pitches to publishers.
“In order for an artist to understand exactly what they’d sound like doing a song, some songwriters are now using Ai voice generation to better tailor the song pitch. While the ethics of this might be in question to some, there hasn’t been much pushback for the idea so far, since the Ai-generated soundalike isn’t released commercially.” — Hypebot
IMPORTANT: In the event that the “first choice” copyright owners rejected your demo, you would still retain full rights to your song composition and have the option to replace the original AI clone voice with the voice of another artist and then approach (shop your track) to other artists/ownership rights holders.
“But, to me, I don’t see this as a threat. A lot of people are freaking out at the moment. I just see this as another tool for us to make better records, make better demos. If I would wanna sell a record to Ariana Grande [or] Rihanna, I think it’s a good way to have her singing on it, so that she knows exactly how it’s gonna sound.” — David Guetta, DJ with over 14 Billion streams
Automated upload options to distribute the new track to selected online music destinations and platforms (Spotify, Prime, Apple Music, Tik Tok, YouTube, etc). Also note, depending on the arrangements and contracts setup with larger streaming platforms, it may initially make sense to “go direct” and stream tracks directly off the platform to subscribers at least until mainstream acceptance of AI vocal songs is deemed viable.
Full suite of digital marketing, social media resources and, promotion tools via the Digital Audio Workstation (DAW).
Let’s Make an AI Voice Clone Song
Let’s go digging in the digital crates. Let’s say I am an aspiring music creator/producer and I’m looking for inspiration from an old song that captures the spirit of my next project. Let’s also pretend that I am a big Bob Marley and Reggae music fan and I want to create a new anti-establishment song that could become an anti-war anthem. I have written some drafts of songs that are pro-peace, and anti-war but as of yet, don’t have a magical track.
I log on to the Next Spotify and do an AI assisted search through hundreds of thousands of songs uploaded for listening and immediate licensing by independent record labels that have made their tracks available for collaboration deals on the platform. The criteria in my prompt include a reggae-style tempo, lyrics about peace versus war, and a positive overall vibe to the message or “theme” of the song.
The AI returns a list of seventy-two curated tracks, organized by the closest level of relevance to my initial prompt criteria. I listen to several tracks and fall in love with the fifth song listed called “General of Peace” by a band called Beat the Clock. The song was released on an album entitled “Funk Bus” in 1992.
Once I click accept, Booty Rock Records, the record label which owns the license rights to the track, is notified that General of Peace has been selected for a proposed mash-up using Bob Marley’s AI-generated voice. Final negotiations for a percentage of royalties for a reimagining of their song begin only if/when the licensed talent (Marley’s license holder) approves a demo and enters into a contract.
Now I will use the lab portion of the Next Spotify platform to create the new updated version of “General of Peace” (my reimagined vision) using AI tools and related production software. Using the original track as the foundation, I create a deeper bassline, stripped down guitar riffs, and add some reggae-style horns. Then I add the AI vocals. After several attempts, the AI voice generation technology provides a near-perfect rendition of Bob Marley singing “General of Peace” in the uptempo, ska music style I’m seeking. I now have a well-produced (but maybe not perfect) demo track to “pitch” Primary Wave, the license holders for the music of Bob Marley, (again, this is completely hypothetical).
The brand new “General of Peace” is put into a waiting queue (the pen) until a representative from Primary Wave has time to log in to MixMashter to review the track and decide if they are interested in pursuing a co-licensing deal. Marley’s Estate and Primary Wave love the track and agree to enter into an agreement with plans to release the song as part of a new anti-war concept album comprised of new songs utilizing Bob’s AI-voice-generated vocals. While they recognize the production is decent, Primary Wave recommends re-recording the foundation track with a live bass player and adding some female background singer vocals for the chorus. That ball is in their court.
Sure! Nooooooo problem!
“If embraced, AI could unearth an entirely new set of licensing opportunities for recorded music, from licensing proprietary datasets to consumer-creation tools.” — Mark Mulligan, MIDiA
Negotiations (again…hypothetical)
Flat one-time license fees paid to the record label and the new song creator are an option, but let’s look at a scenario where all stakeholders receive a percentage share of a potential breakout song.
Because of the value of Bob Marley’s infamous voice, Primary Wave will receive the lion’s share of revenue generated from the song (65%). In this case, Primary Wave opts to control all promotion, packaging, and distribution of the track, (risk assumed) hence the 65% percentage rather than 80–85%.
The independent record label and license owner of “General of Peace” agrees to receive 15 % of the royalties generated for the use of their copyrighted material. This factor may become highly relevant as AI copyright law evolves. The fact that the lyrics were originally written by a human, and reimagined/co-produced by a human using the AI vocals of consenting ownership partners (Primary Wave) could make or break the track from being accptable for sale/streaming in the eyes of the RIAA and major streaming platform policies.
The Next Spotify (the AI tech and licensing stack platform) receives 10% royalties for services provided. Note: Primary Wave might have the option to “earn back “ that equity ownership via incremental royalty payments made to MixMashter over time if the song becomes a mainstream hit.
Me, the creator who developed the idea, reimagined the song and is the producer who shopped the track receives 10% of the royalties. While this may seem paltry, I have now got my foot in the door of the music industry. The fact that I successfully ideated, sourced, and produced the “demo” for a great track with wide distribution from an A-List artist will bring me clout and more negotiation power with other rights holders in the future, particularly with less famous bands and singers and I will be able to negotiate larger royalty percentages or buyouts, etc.
Please note, there is the possibility of utilizing smart contracts and the blockchain to streamline and easily solidify contracts. Also, the vast potential for using NFT’s as promotional tools and possibly even towards raising capital for new collaborative projects is fascinating, but we are going to let that sleeping dog lie…for now.
Run It Back
Let’s go digging in the digital crates once more. Except this time, we use AI-assisted lyrics combined with your own original lyrics. You decide to use the DAW to spin up a concept demo you have been noodling on for a week. You enter a prompt into the ChatGPT style interface with an LLM trained specifically on Hip Hop music.
Your prompt: Create a “rap posse” song similar to “The Symphony” by DJ Marley Marl using your original rhymes. Replace the four rappers Masta Ace, Craig G, Kool G. Rap, and Big Daddy Kane with the voices of Tupac Shakur, Notorious B.I.G., Guru, and AdRock from the Beastie Boys. Use the original underlying loop from “Hard to Handle” by Otis Redding combined with your lyrics for each rapper with selected samples from Dr. Dre.
Then on a whim…
Your next prompt: “Create a song using the rock band Queen with Freddie Mercury lead vocals singing “What’s New Pussycat?” by Tom Jones over the instrumental version. Then add some Shirley Bassey background vocals and make the production quality sound like a cross between a John Barry “James Bond” movie theme and the swinging style of early Quincy Jones.
You get the results for both prompts. Your “Dead Poet’s Society” rap sequel to the Symphony is D.O.A. It lacks the tempo and the energy of the original masterpiece by Marley Marl and needs some (a lot) of work. However, the track of Freddie Mercury singing “What’s New Pussycat?” sounds incredible! Somehow capturing the retro sound of the bygone era and yet feeling relevant right now. You’ve caught lightning in a bottle! Odd, because you were almost certain your rap posse song would be great and Pussycat was at best, a long shot.
Now begins the re-editing, mastering, and clean-up of your red-hot demo track before submitting it for approval on the Next Spotify music platform.
“The concept at the core of why creative AI could be so disruptive is that the friction between having an idea, and manifesting that idea in the form of a creative work, has essentially gone down to zero.” — Cherie Hu, Founder, Water & Music
Note: There is a lot going on here and obviously a need to be sensitive to the concept of (for lack of a better term(s)) digital blackface / cultural appropriation. In this context referring to the inevitable droves of non-black songwriters and rappers that will write rhymes for current or deceased and current black artists, utilizing their voice clone combined with their own original lyrics and beats. This is an important topic that will need to be discussed as the industry moves forward.
AI voice clone songs should require the use of a unique tag that identifies the song as AI-assisted. Deezer claims it is developing technology that will identify AI generated content on its platform. Going forward, such technology would be highly beneficial to large music distribution platforms if indeed, a new AI voice clone music vertical is created and could be categorized as such on the major streaming platforms (as Grimes has suggested). Additionally, information about the songwriters and production might also be included allowing the listener/customer to evaluate, consider and determine how they feel about each track.
“As the most convincing attempts would suggest, AI works best when it can draw on the analysis of extensive back catalogues, when it involves fairly heavy human intervention, or when it uses singers who have a definitive style that can be pitched or blurred to cover up for any clunkiness.” — Jenessa Williams, The Fourty-Five
The Last Beatles Song — The Third Signpost
The release of the new Beatles track might be a monumental moment in music history. The song, titled “Now and Then” was recently completed using AI to clone the voice of John Lennon and will be released later this year. It should be noted, that the AI was used specifically for the restoration of Lennon’s voice due to the degradation of the cassette tape that the song was originally recorded. This was a wonderful announcement from one of the greatest bands ever. Perhaps more importantly, the release could be a watershed moment in music history and may represent the first stage of mainstream music fans around the world grasping the very personal and unquantifiable value of vocal clone songs from artists past and present.
Reaction time is a factor, and the industry should keep a very close eye on how music fans and the general public embrace or reject the AI-assisted Beatles song. Personally, I think there will be an outpouring of love and appreciation for the magnificence of the Beatles band and renewed interest in their music catalog. While some will remain purists and dismiss AI music tech, I believe many others will over time, accept and look forward to new hybrid AI music tracks and albums from their favorite legacy bands and singers, and the Beatles AI assisted song could be the key. Part loving tribute, part homage, part musical novelty, AI voice-cloned songs from legendary acts may be controversial but will undoubtedly create an insane amount of industry buzz leading up to release dates.
You just can’t tell me that a new AI vocal clone David Bowie or Aretha Franklin song won’t also spur massive interest, which would in turn, create a spike in streaming subscriptions as well as vinyl and CD sales at the few remaining record stores (that would be eternally grateful for the renewed interest in potentially thousands of older, fading artists). So if I’m a record label executive, who according to Barron’s, is currently struggling with a decline in ad sales and growing concerns about a slowing growth rate of streaming music subscriptions, while simultaneously sitting on a potential wealth of assets from legacy artists that are eroding due to my primary demographic dying off, this AI vocal clone thing should be music to my friggin’ ears.
The industry needs to strongly consider about twelve well-planned, well-produced new AI songs released each year from the Beatles, Bob Marley, Nirvana, Selena, Jimi Hendrix and many others using AI voice cloning technology (they used to call these things “albums”). Notice a pattern? All of these artists’ catalogs were cut short for various reasons, leaving tens of millions of their devoted fans feeling shortchanged and hungry for more.
Why only twelve AI-assisted songs per artist released each year? Because we are still in the early stages of AI-generated music and we need quality over quantity. And it goes without saying, there are plenty of ethical and legal issues that still need to be sorted out on the business side and the copyright owners and record labels will need time to properly plan, produce and promote the release of AI tracks from their iconic bands and singers to a global fanbase who is also processing the ethics of vocal clone technology themselves.
A Celebration of Life and Musical Genius
So my proposal is that on the annual anniversary date of the release of the last Beatles song, a new “Reimagined Beatles” album would be released with 12 tracks. Eight of the twelve songs would be written by the top songwriters selected by McCartney, Ringo Starr, and close friends and family (with some input from representatives from Sony Music), produced using the best musicians and studio technicians combined with the AI voices of the Beatles. Two tracks written and created by aspiring producer Beatles superfans (again with input from McCartney and Sony). Two songs written/created by AI (ChatGPT, Bard style derived from a dedicated Beatles LLM) and performed by live musicians and singers that are “voted in” to the album by fans online similar to the way players are voted to play in the NBA All-Star Game.
On the next day and the rest of “Reimagine” week, new AI-assisted albums would be released from other legendary acts including The Doors, Pink Floyd, The Beach Boys, Curtis Mayfield, Rick James, Aretha Franklin, Ella Fitzgerald, Mel Tillis, Howlin’ Wolf, Johnny Cash, Roy Orbison, Selena, and many others in the same general format as the Beatles Reimagined album. Each release would be produced using top songwriters combined with world-class musicians and production teams. Turn each Reimagined release into a celebration of love and appreciation of fans around the world for their favorite all-time greats.
A short leash on major AI song releases may sound like it will cut out thousands of talented independent producers, but on the contrary, it will open the door to new opportunities, provided that there is a platform in place that can provide all of the services necessary for indie producers to participate in the AI vocal clone song revolution. In fact, combining the voices of deceased artists with modern songs and remixes is actually nothing new. Jazz labels Blue Note, Verve and others have created remixes and original tracks using the voices of Sarah Vaughn, Ella Fitzgerald, and many, many others.
What’s At Stake
Estimates from Goldman Sachs for the entire global music industry are $92 billion in 2023 and are forecast to reach approximately $151 billion by 2030.
Variety reports that is 2022, “the U.S. recorded-music industry reached an all-time high of $15.9 billion in 2022, marking the industry’s seventh consecutive year of growth.”
Streaming: Grand View Research predicts that “the global music streaming market size was valued at USD 29.45 billion in 2021 and is expected to expand at a compound annual growth rate (CAGR) of 14.7% from 2022 to 2030.”
Goldman Sachs indicates that Spotify is the industry leader, currently capturing approximately 34.4% of global market share with Tencent and YouTube making gains while Apple Music and Amazon Prime Music remained stable.
Meanwhile, according to Verified Market Research, the US music streaming service market size was valued at USD 29.70 Billion in 2021 and is projected to reach USD 118.03 Billion by 2030, growing at a CAGR of 16.96% from 2023 to 2030.
According to DemandSage, Spotify is the world’s leading audio streaming service, with 515 million users and 210 million premium subscribers across 184 regions. Over 44% of Spotify subscribers use it everyday.
Zippia.com reports that “globally, there are 616.2 million people who have paid music streaming subscriptions. This was the number as of the end of 2022, and it was up by 93 million people from 2021. At the end of 2022, there were 82.2 million people with paid streaming subscriptions in the U.S. alone.”
“ Napster pays the most per stream of any streaming service at $0.019 per play. TIDAL comes next with $0.0125 per stream, and Apple Music takes third place, paying $0.00783. Spotify, the largest streaming platform in the world, is near the bottom of the list as far as what it pays artists, giving them between $0.003 and $0.005 per stream.” — Zippia.com
2022 industry stats from Zippia.com include:
- The annual revenue of the global recorded music industry is $26.2 billion as of 2022.
- The U.S. is the top music market in 2022, followed by Japan, the U.K., and Germany.
Some relevant 2020 stats from Zippia:
- Universal Music Group is the world’s largest record company, earning 32.1% of the global streaming and physical music industry’s revenue in 2020.
- Independent record labels accounted for 31.4% of the world’s streaming and physical music sales revenue in 2020.
- Sony Music Entertainment earned 20.6% of the global revenue earned from streaming and physical music sales in 2020.
- 15.9% of the revenue earned by the world’s physical and streaming music sales in 2020 went to Warner Music Group.
Source: Zippia.com
Unfortunately for the majority of musicians, streaming revenue just doesn’t pay the bills. While a select few Mega star A-List artists (Beyonce, Drake, etc) capture the lion’s share of the total number of streams, the majority of artists garner just a small fraction of the estimated 12% going to creators. I will just say it, all this nonsense and doublespeak about losing the “human connection” and “emotion in the music” is mostly a smokescreen for a terrified industry circling the wagons as they scramble to figure out how not to get disrupted again and lose a portion of their obscene profits.
“The company’s business model, to put it simply, is built on the severely underpaid labor of millions of creators. “ — Otherwords
Concerning AI music forecasts, there is limited research. However, according to Market.Us “the global Generative AI in Music Market size accounted for USD 229 million in 2022 and growth is estimated to accelerate at a CAGR of 28.6%, registering an incremental revenue of USD 2.6 billion by 2032.”
According to Play.ht, “Market Reports World says the global voice cloning market is experiencing substantial growth. It was worth $461.6 million in 2022. At an annual growth rate of 24.6%, it’s expected to expand and reach $1,723.9 million by 2028.”
I believe the AI music category could see a significant spike if voice cloning songs were to catch on with mainstream audiences. But there are challenges: Will older audiences be reinvigorated by hearing new songs from some of their favorite classic artists? Will this lead to more subscriptions and retail sales? What about revenue from advertising? Will younger audiences be turned off if their favorite rapper is spitting lyrics alongside a legendary but deceased OG? Will global fans become more engaged in their personal music discovery journey if there is huge buzz from controversial but exciting changes from AI in the industry?
We’re going to find out if AI will move the needle in the music industry over the next year.
“Labels now face challenges like slowing global revenue growth in recorded music, particularly in established markets like the U.S, while the emergence of AI-generated music adds to their concerns, as many laws do not currently deem the replication of artists’ voices as strictly illegal.” — Reuters
Label Silos
Artificial Intelligence presents a massive opportunity for innovative entrepreneurs and artists to disrupt current industry stalwarts and participate in the AI music revolution. But will the recording industry and mainstream artists circle their wagons and create label silos, locking down all of their assets from any sort of AI experimentation while they buy time to build out their own AI infrastructure while simultaneously trying to squash the AI music movement over questionable copyright issues? Or will they join ranks with the vast armies of indie AI music producers, capitalizing on the massive paradigm shift from passive listening consumption to active production participation, and provide some of their assets as the genesis of an explosive growth segment in the music industry?
The answer is actually both. Silos in the short term to stunt the onslaught of training the Large Language Models using copyrighted material while the majors scramble to figure things out and then over time, morph into a hybrid form of “open source” collaboration once a viable production, distribution and e-commerce model (the Next Spotify platform) is in place.
“By restricting access to these catalogs, the training of these models is slowed down, thereby postponing the moment when these tools become effective enough to pose a significant threat to the actual majors’ businesses. This approach provides valuable time for the industry to prepare for the technological evolution and adapt their business models and strategies accordingly.” — Honore Piffaut AI-GENERATED MUSIC — Navigating Disruption : How Generative AI could affect the sustainability of major players in the music industry?
Spotify Feels the Pressure
In July, Spotify finally caved to Wall Street and raised its subscription prices.
“In the U.S., Spotify’s Premium Individual offering now costs $10.99, up from $9.99, and the price of its Premium Duo plan changed to $14.99, up from $12.99. The company’s Premium Family plan is now priced at $16.99, up from $15.99, and the Student offering costs $5.99, up from $4.99.” — CNBC
According to the New York Post, this price bump is in addition to cutting 800 jobs in January and cutting an additional 200 more in June. If it’s true that the Spotify and presumably other streaming platforms are feeling pressure from TikTok’s new music streaming service, revenue shortfalls and are having to lop heads and raise subscription prices to meet Wall Street expectations, don’t they need a shot in the arm? Something truly mindblowing that combines real, heartfelt nostalgia with groundbreaking innovation? A unique feature or aspect to their service that can seperate them from their streaming competitors? A…..wait for it….DIFFERENTIATOR?
The Platform is the Product
Music has been commodified. The ritual of going to a record store, sifting through the new releases and rushing home with a new album and a couple of singles is ancient history (for most). The platform is now the product (Spotify, Apple Music). A new platform that released into the wild that offered nothing new will not make it. However, a service that provided exclusivity to AI tracks might be unique and novel enough to scale globally as a dedicated category within an existing platform but also, depending on the level of acceptance and growth of AI vocal clone songs, could be a stand alone service. While Spotify now has over 100 million songs to choose from, so do all of the other major streaming platforms like Amazon, Google, Apple, Deezer, Tidal, Youtube, and others. AI collaboration songs would represent “original content” which could be made exclusive to the new platform. Initially, it may be that only smaller and middle-sized record labels agree to license the voices and tracks of their bands to be reimagined by aspiring artists and composers using AI production tools. But that still represents a huge opportunity.
What if there was a new platform that published only AI-exclusive tracks that were fully copyright protected through automated contract agreements between all vested parties prior to distribution into the marketplace?
What if music fans could only stream the exciting new collaborative AI tracks via paid subscription from that new platform?
What if all of the tracks offered were available in any language you wanted? Don’t forget, AI can also quickly swap out languages so that a catchy track could be immediately translated and distributed to other countries in their native language once a clear “hit” had been created and gone viral domestically.
What happens when you have tens of thousands of aspiring songwriters, bands and producers with access to professional quality recording equipment, now coupled with access to the iconic voices of the greatest singers in music history, all producing exclusive tracks for this new platform and dropping exciting new remixes, reimagines and mash-ups every day?
Enter The Drop Spot
The Drop Spot will be the secret sauce for the Next Spotify if it is built as a standalone streaming platform. Whoever wrangles with the major labels and independents MUST figure out a way to gain exclusivity to new AI voice clone songs for (at least) the first week of their release. All new AI song drops would be made available for streaming in the Drop Spot for an exclusive time period which would justify an additional fee above and beyond the subscription price to access the same existing back catalog of songs that everyone also has access to. If the Next Spotify AI voice clone concept is integrated into an existing music streaming platform it would at a minimum be a key product differentiator and at best, would allow for a higher subscription tier for serious fans to gain immediate access to new AI voice clone song drops.
The Sound of Inevitability
While there will be debate over ethical issues and ownership of the personal brand, etc, the voices of mega-star pop singers will increasingly become recognized as highly valuable assets under the control of record labels. Stakeholders may come to view the potential revenue of the more iconic voice assets as significant “found money”, leading to additional pressure to capitalize using AI.
Mainstream audiences will be exposed to more black-market AI songs and begin to accept voice cloning of their favorite artists. The pressure on the three major US record labels to at least experiment will mount. For Direct Service Providers (DSP’s) like Spotify, pressure may be exacerbated by market forces including slowing streaming subscription revenue, Wall Street pressuring to raise subscription prices, and the subsequent need to create “added value” through something fresh and exciting like new releases from new and old favorites using AI voice clone tech.
As opportunities for remixing, reimagining, and creating mash-ups arise using their voice assets, independent record labels and rights owners will begin entering into AI voice cloning projects. Most likely, this will begin with fully independent artists (Grimes) and then expand to small to mid-size record labels dipping their toes with select portions of their catalogs. Upon market acceptance, the mid-size independent labels will go all-in on voice cloning tech and will either create or (most likely) partner with tech companies offering Large Language Models (LLM)’s to train generative AI models using their entire catalogs.
Once there is a consecutive string of voice clone hits and the model is considered viable, the three large record labels will enter the fray, leading to widescale licensing opportunities of their existing voice assets and the potential for the creation of new smash hits by aspiring producers songwriters and producers. While its possible the record labels will offer their own private voice cloning tech platforms (Label Silos) for quality control purposes, it may make more sense to simply utilize a third party who manages all of the software development, tech scalability issues, legal service automation, sales and customer service to reduce expenses and headache.
Unfortunately, there are going to be losers in this game and plenty of economic pain to go around as large swaths of hard-working professional composers, technicians, aspiring singers and session musicians now face the real possibility of being made irrelevant due to groundbreaking automation technology from Silicon Valley. And I doubt they get a lifeline from the record labels. Based on the cutthroat nature of industry history, I think the major labels will actually accelerate this process as they shrewdly cut out studio musicians and production teams in favor of the sizable cost savings of AI music automation tools.
Spotify and other large platform players are not immune either, as the AI music revolution could shift a significant portion of visitor traffic from sites simply offering music aggregation and distribution to more advanced platforms providing subscribers with full-scale AI production tools, automated copyright and legal services, social profiles/brand pages, audio, video editing suites, libraries of content available for collaboration and the meticulous curation of the best tracks from the millions of new AI clone songs that will be generated weekly on the Next Spotify platform.
Needless to say, vast fortunes will be made, lifelong careers will be altered, wild west dealmaking and music industry business disruption will commence daily at a relentless pace for the foreseeable future. And for billions of music fans around the world, the discovery of the resurrected musical genius of thousands of talented but long-forgotten songwriters will become a daily occurrence.
Buckle up
“Artists should see AI as an opportunity, and extension of their brands. Future music fans are going to design the music that they want to hear. The creative class must learn to adopt AI and leverage their skills effectively, not lament the inevitable.” — The Hill
Peer into the Crystal Ball
AI might also revolutionize podcasts, audiobooks and perhaps even radio broadcasting. Imagine selecting an AI play-by-play throwback broadcast by Harry Caray for a Cubs baseball broadcast. Imagine listening to an audiobook that adds background sounds as the antagonist traverses through the narrative of a 1940’s murder mystery. “He enters a busy restaurant in Chinatown, seeking a clue why she made a detour before heading home to her apartment in mid-town” (with the sounds of footsteps descending down concrete steps, a door opening, patrons eating, ordering food, busy kitchen noises, plates clattering, and patrons chatter in the background). All sounds created using AI text to voice/soundscape prompts layered into the AI voiceover narrative read by a Humphrey Bogart voice clone.
Lastly and one of the most compelling reasons AI voice clones of legendary bands and artists will become accepted in the marketplace in the future is that AI technology will be a central component of virtual reality (VR) and augmented reality (AR). AI vocal cloning will be used in VR concerts and performances by current and past bands and singers, including deceased artists where the copyright owners and estates have entered into agreements with large VR/AR technology companies.
Imagine logging into an AR/VR headset for a three-song acoustic performance by Tracy Chapman in your living room where Tracy engages with the viewer through voice prompts, shares a personal story between songs, and verbally responds to requests for favorites, etc. This might be followed by switching from AR to VR to step back in time for an experiential two-hour concert of The Doors at the Whiskey a GoGo on the Sunset Strip, circa 1966 Los Angeles while sipping a brew (real) and engaging in conversations with members of the Mamas and the Papas whose AI personas are also virtually in attendance. This type of tech isn’t fantasy, it’s already here. It represents the digital side of the experience economy ( probably the most lucrative) and (IMO) the future of the AI-assisted music/entertainment industry.
And you thought Hot Tub Time Machines were a crock of shit.
A Few Observations
Voice generators are here to stay and will become a standard publishing tool for voice reclamation of deceased artists, aging artists, and most importantly for artists too busy to sing or provide voice-over for low-budget or start-up music projects. Savvy industry vets and recording artists are already building AI solutions and investing in AI music technologies.
Quality control of AI-generated tracks featuring the AI-generated voices of major mainstream artists will be critical. The major labels will closely guard their breadwinner talent but may begin to experiment with under-the-radar releases from their more obscure or less relevant talent initially to see what works and what doesn’t.
A new type of music executive will emerge, the CDO (Chief Discernment Officer) will emerge that (smashing pu mpkins article here_ has a special ear for knowing what types of songs will work with voice gnerators of certain artists. Sony Music recently made a move to get out in front of AI with the addition of a new AI czar.
AI will never replace the intimacy of a live performance. Musicians have a very special bond with their fans and audiences at live concerts that recorded music can’t match. This will never change.
HOWEVER. what may change is the way fans attend live events. AI-assisted virtual reality concerts will be an area to watch closely over the next couple of years. AI-assisted VR technology may allow tens of thousands of people to virtually attend an intimate concert by Alicia Keys at a small, sold-out venue. Further out, it could also allow people to have front-row seating at a Michael Jackson concert or watch The Doors perform at the Whiskey GoGo in 1966 Los Angeles. Virtual concerts on Roblox are a start, but as AI applications evolve, the tech might bring these surreal VR experiences to a whole new level. As it relates to voice cloning, imagine the hype of being on a VIP list for new AI-assisted song drops by current and past performers in VR.
Some music categories seem better suited for disruption by AI, particularly, Hip Hop, House, Techno, EDM and most Pop music. This is because the majority of the production of these categories is less reliant on traditional instruments and instead, uses advanced music software, excluding the vocals in Hip Hop and some R&B and mainstream Pop as well.
Music genres that seem well suited for AI experimentation include:
Blasts from the past — Older artists who have faded but still have a loyal fanbase. Examples from the rap industry, think Slick Rick, Doug E Fresh, Big Daddy Kane, Salt n Pepa, Special Ed, The Fat Boys, Biz Markie, etc, etc. Brand new jams that suddenly sound vibrant and exciting with classic rappers spittin’ new rhymes over bad-ass beats and samples.
Parody and Bizarre Mash ups — Biden and Trump raps come to mind.
Gone but not forgotten — David Bowie, Prince, Freddie Mercury, Jim Morrison, Janis Joplin, Jimi Hendrix, Bob Marley, Aretha Franklin, Curtis Mayfield, Teddy Pendergrass, and on and on and on.
“Generative AI is groundbreaking in that it allows content creators to generate unique musical compositions without the need for musical experience or expensive equipment. In 2021, a project known the Lost Tapes of the 27 Club made headlines when the project used AI to create “new” tracks by artists who lost their lives to suicide at the age of 27, including Kurt Cobain, Jim Morrison, Jimi Hendrix, and Amy Winehouse.” — Albert J. Soler, Scarinci Hollenbeck
How did we get here?
Was Heart on My Sleeve really a Napster moment for the record industry, or was it actually an MTV moment? Something that is blatantly obvious and represents a massive, seismic opportunity for the record labels and music industry but is initially rejected and frowned on because it is a radical departure from the status quo?
Believe it or not, investor feedback to pitches by MTV frontman Bob Pittman was initially cold and unresponsive. Music videos on television? Bah humbug! Rubbish! Pure Nonsense! People listen to music on the radio, they won’t “watch music” on television! The pushback sounds absurd now, but watch the “I want my MTV” documentary, the struggle was real.
Let’s do a quick review of the evolution of music technology and its impact on society, and the way people discovered and listened to new music.
The vinyl record album(wax)
8 Track (essentially designed for listening to favorite albums/artists in cars)
The tape cassette made music truly portable and was the first technology to usher in “music piracy” via the recording of live radio broadcasts and also the creation of curated “mixtapes”. Mixtapes were made by recording selected songs from vinyl albums by connecting a home phonograph to a tape deck recorder (the original “ripping”). I would argue they had a huge impact on the explosive growth of punk, new wave, and early hip hop in the late seventies and early eighties.
Then the release of the Sony Walkman (and dozens of knockoff headsets) to play personalized mixtapes, pre-recorded albums, and (towards the waning days of the format) ridiculously overpriced “tape singles”.
The “Boom Box” also allowed for music portability with radio and cassette and became an iconic cultural artifact in the late seventies and early eighties. This was also an “under the radar” method of music discovery.
The introduction of the compact discs (CD) and most importantly, the switch to digital. This represented another massive money grab for the recording industry and the major labels made an absolute killing from millions of music fans switching to the new digital format, paying roughly $15 for a cd to replace their old, scratched vinyl albums. Well, at least CD’s make for fun cocktail coasters now.
Let it Rip!
The conversion to digital music kicks off the college pastime of ripping cd’s to MP3 format
Apple Music and the iPod, the incredible hardware product and digital update to the portable Sony Walkman, were later replaced by the even more incredible iPhone. Personal Note: It’s impossible to relay how badass the original iPod was for music lovers.
MP3.com and the slippery slope to……
Napster.
The record industry loses its shit as their sales start tanking leading to the demise of hundreds of traditional record stores as we knew them. Nobody talks about these people being displaced but it happened. And it happened pretty fast.
It wasn’t until another tech company (Apple) came up with iTunes as a safe and simple alternative that swayed mainstream America to reconsider paying for music again.
Unfortunately for the record industry and many artists, the behavior of consumers radically changed. No longer were they going to shell out $16.00 for a cd. Now they could pick and choose the exact songs they wanted for a buck a piece on iTunes. Further, the majority of younger consumers who had grown up with Napster were even more jaded and were only lured back to “paying” for their music consumption by putting up with some advertising during streams (YouTube) or paying a small monthly subscription fee (Spotify — and in many cases, still putting up with ads). The music industry had no choice. A classic case of “if you can’t beat ’em, join ’em.
The rampant copyright infringement goes supernova and the RIAA cracks down hard on Napster, leading to innovation and early streaming services including Pandora Radio.
The early social network MySpace becomes a popular online destination for discovering music before Facebook emerges with a better (more uniform) UX, that allows for mainstream social media adoption, and YouTube (snapped up by Google) emerges to offer an unbeatable online platform for music discovery and viewing music videos, simultaneously taking out an already fading MTV for music video discovery.
Spotify officially launches in the US in 2011. Created by Swedish businessmen, Daniel Elk and Martin Lorentzon. The platform is subscription-based and ad-supported and allows users to stream music on demand. Other music streaming services include Apple Music and Prime Music. Juggernaut TikTok emerges as a music discovery tour de force through rampant user-generated content. Note, TikTok announced the launch of its own streaming platform based on its #NewMusic search hub.
Meanwhile Fruity Loops, Pro Tools and Garage Band software tools emerge as the go to tools for fast and easy songwriting and music production from home.
Guitar Center brings instruments, concert gear and software to the masses.
So, why the history lesson? The reason is that in order to know what comes next, you should know what came before. So in this case, the transition to ripping files and peer-to-peer sharing by Napster is relevant because that event put the power in the hands of the consumer/end user. Fast forward twenty-plus years to now and AI puts even more power in the hands of the end user in the form of easy production tools with insane quality (near-perfect renditions of singers’ voices) and the ability to distribute new tracks around the world via social network platforms. There are now no barriers.
FIN
Music is the one unifying force of all the people across the entire planet. When someone creates a beautiful song and puts it out in the world, is it still all hers? Or to a certain extent, is it now part of the collective consciousness shared by the millions of people who adore it and make their own personal connection with it? There are few (if any) forms of creative expression that can match the unbridled passion experienced by the fan/listener. Is there a way to democratize music for everyone to create, listen and enjoy together and bridge generations through their love and appreciation for the magical quality of legendary artists?
UPDATE: A major step forward for AI music generation and voice cloning announced by YouTube and DeepMind (11/16/2023)
