Minds, machines, and centralization: AI and music

Far from the liberated playground the Internet once promised, online connectivity now threatens to give us mainly pre-programmed culture. As we continue reflections on AI from CTM Festival in Berlin, here’s an essay from this year’s program.

If you attended Berlin’s festival this year, you got this essay I wrote – along with a lot of compelling writing from other thinkers – in a printed book in the catalog. I asked for permission from CTM Festival to reprint it here for those who didn’t get to join us earlier this year. I’m going to actually resist the temptation to edit it (apart from bringing it back to CDM-style American English spellings), even though a lot has happened in this field even since I wrote it at the end of December. But I’m curious to get your thoughts.

I also was lucky enough to get to program a series of talks for CTM Festival, which we made available in video form with commentary earlier this week, also with CTM’s help:
A look at AI’s strange and dystopian future for art, music, and society

The complete set of talks from CTM 2018 are now available on SoundCloud. It’s a pleasure to get to work with a festival that not only has a rich and challenging program of music and art, but serves as a platform for ideas, debate, and discourse, too. (Speaking of which, greetings from another European festival that commits to that – SONAR, in Barcelona.)

The image used for this article is an artwork by Memo Akten, used with permission, as suggested by curator and CTM 2018 guest speaker Estela Oliva. It’s called “Inception,” and I think is a perfect example of how artists can make these technologies expressive and transcendent, amplifying their flaws into something uniquely human.

Minds, Machines, and Centralisation: Why Musicians Need to Hack AI Now

IN THIS ARTICLE, CTM HACKLAB DIRECTOR PETER KIRN PROVIDES A BRIEF HISTORY OF THE CO-OPTING OF MUSIC AND LISTENING BY CENTRALIZED INDUSTRY AND CORPORATIONS, IDENTIFYING MUZAK AS A PRECURSOR TO THE USE OF ARTIFICIAL INTELLIGENCE FOR “PRE-PROGRAMMED CULTURE.” HE GOES ON TO DISCUSS PRODUCTIVE WAYS FOR THOSE WHO VALUE “CHOICE AND SURPRISE” TO REACT TO AND INTERACT WITH TECHNOLOGIES LIKE THESE THAT GROW MORE INESCAPABLE BY THE DAY.

It’s now a defunct entity, but “Muzak,” the company that provided background music, was once everywhere. Its management saw to it that their sonic product was ubiquitous, intrusive, and even engineered to impact behavior — and so the word Muzak became synonymous with all that was hated and insipid in manufactured culture.

Anachronistic as it may seem now, Muzak was a sign of how tele-communications technology would shape cultural consumption. Muzak may be known for its sound, but its delivery method is telling. Nearly a hundred years before Spotify, founder Major General George Owen Squier originated the idea of sending music over wires — phone wires, to be fair, but still not far off from where we’re at today. The patent he got for electrical signalling doesn’t mention music, or indeed even sound content. But the Major General was the first successful business founder to prove in practice that electronic distribution of music was the future, one that would take power out of the hands of radio broadcasters and give the delivery company additional power over content. (He also came up with the now-loathed Muzak brand name.)

What we now know as the conventional music industry has its roots in pianola rolls, then in jukeboxes, and finally in radio stations and physical media. Muzak was something different, as it sidestepped the whole structure: playlists were selected by an unseen, centralized corporation, then piped everywhere. You’d hear Muzak in your elevator ride in a department store (hence the phrase, elevator music). There were speakers tucked into potted plants. The White House and NASA at some points subscribed. Anywhere there was silence, it might be replaced with pre-programmed music.

Muzak added to its notoriety by marketing the notion of using its product to boost worker productivity, through a pseudo-scientific regimen it called the “stimulus progression.” And in that, we see a notion that presages today’s app behavior loops and motivators, meant to drive consumption and engagement, ad clicks and app swipes.

Muzak for its part didn’t last forever, with stimulus progression long since debunked, customers preferring licensed music to this mix of original sounds, and newer competitors getting further ahead in the marketplace.

But what about the idea of homogenized, pre-programmed culture delivered by wire, designed for behavior modification? That basic concept seems to be making a comeback.

Automation and Power

“AI” or machine intelligence has been tilted in the present moment to focus on one specific area: the use of self-training algorithms to process large amounts of data. This is a necessity of our times, and it has special value to some of the big technical players who just happen to have competencies in the areas machine learning prefers — lots of servers, top mathematical analysts, and big data sets.

That shift in scale is more or less inescapable, though, in its impact. Radio implies limited channels; limited channels implies human selectors — meet the DJ. The nature of the internet as wide-open for any kind of culture means wide open scale. And it will necessarily involve machines doing some of the sifting, because it’s simply too large to operate otherwise.

There’s danger inherent in this shift. One, users may be lazy, willing to let their preferences be tipped for them rather than face the tyranny of choice alone. Two, the entities that select for them may have agendas of their own. Taken as an aggregate, the upshot could be greater normalization and homogenization, plus the marginalization of anyone whose expression is different, unviable commercially, or out of sync with the classes of people with money and influence. If the dream of the internet as global music community seems in practice to lack real diversity, here’s a clue as to why.

At the same time, this should all sound familiar — the advent of recording and broadcast media brought with it some of the same forces, and that led to the worst bubblegum pop and the most egregious cultural appropriation. Now, we have algorithms and corporate channel editors instead of charts and label execs — and the worries about payola and the eradication of anything radical or different are just as well-placed.

What’s new is that there’s now also a real-time feedback loop between user actions and automated cultural selection (or perhaps even soon, production). Squier’s stimulus progression couldn’t monitor metrics representing the listener. Today’s online tools can. That could blow apart past biases, or it could reinforce them — or it could do a combination of the two.

In any case, it definitely has power. At last year’s CTM hacklab, Cambridge University’s Jason Rentfrow looked at how music tastes could be predictive of personality and even political thought. The connection was timely, as the talk came the same week as Trump assumed the U.S. presidency, his campaign having employed social media analytics to determine how to target and influence voters.

We can no longer separate musical consumption — or other consumption of information and culture — from the data it generates, or from the way that data can be used. We need to be wary of centralized monopolies on that data and its application, and we need to be aware of how these sorts of algorithms reshape choice and remake media. And we might well look for chances to regain our own personal control.

Even if passive consumption may seem to be valuable to corporate players, those players may discover that passivity suffers diminishing returns. Activities like shopping on Amazon, finding dates on Tinder, watching television on Netflix, and, increasingly, music listening, are all experiences that push algorithmic recommendations. But if users begin to follow only those automated recommendations, the suggestions fold back in on themselves, and those tools lose their value. We’re left with a colorless growing detritus of our own histories and the larger world’s. (Just ask someone who gave up on those Tinder dates or went to friends because they couldn’t work out the next TV show to binge-watch.)

There’s also clearly a social value to human recommendations — expert and friend alike. But there’s a third way: use machines to augment humans, rather than diminish them, and open the tools to creative use, not only automation.

Music is already reaping benefits of data training’s power in new contexts. By applying machine learning to identifying human gestures, Rebecca Fiebrink has found a new way to make gestural interfaces for music smarter and more accessible. Audio software companies are now using machine learning as a new approach to manipulating sound material in cases where traditional DSP tools are limited. What’s significant about this work is that it makes these tools meaningful in active creation rather than passive consumption.

AI, back in user hands

Machine learning techniques will continue to expand as tools by which the companies mining big data make sense of their resources — from ore into product. It’s in turn how they’ll see us, and how we’ll see ourselves.

We can’t simply opt out, because those tools will shape the world around us with or without our personal participation, and because the breadth of available data demands their use. What we can do is to better understand how they work and reassert our own agency.

When people are literate in what these technologies are and how they work, they can make more informed decisions in their own lives and in the larger society. They can also use and abuse these tools themselves, without relying on magical corporate products to do it for them.

Abuse itself has special value. Music and art are fields in which these machine techniques can and do bring new discoveries. There’s a reason Google has invested in these areas — because artists very often can speculate on possibilities and find creative potential. Artists lead.

The public seems to respond to rough edges and flaws, too. In the 60s, when researcher Joseph Weizenbaum attempted to parody a psychotherapist with crude language pattern matching in his program, ELIZA, he was surprised when users started to tell the program their darkest secrets and imagine understanding that wasn’t there. The crudeness of Markov chains as predictive text tool — they were developed for analyzing Pushkin statistics and not generating language, after all — has given rise to breeds of poetry based on their very weirdness. When Google’s style transfer technique was applied using a database of dog images, the bizarre, unnatural images that warped photos into dogs went viral online. Since then, Google has made vastly more sophisticated techniques that apply realistic painterly effects and… well, it seems that’s attracted only a fraction of the interest that the dog images did.

Maybe there’s something even more fundamental at work. Corporate culture dictates predictability and centralized value. The artist does just the opposite, capitalizing on surprise. It’s in the interest of artists if these technologies can be broken. Muzak represents what happens to aesthetics when centralized control and corporate values win out — but it’s as much the widespread public hatred that’s the major cautionary tale. The values of surprise and choice win out, not just as abstract concepts but also as real personal preferences.

We once feared that robotics would eliminate jobs; the very word is derived (by Czech writer Karel Čapek’s brother Joseph) from the word for slave. Yet in the end, robotic technology has extended human capability. It has brought us as far as space and taken us through Logo and its Turtle, even taught generations of kids math, geometry, logic, and creative thinking through code.

We seem to be at a similar fork in the road with machine learning. These tools can serve the interests of corporate control and passive consumption, optimised only for lazy consumption that extracts value from its human users. Or, we can abuse and misuse the tools, take them apart and put them back together again, apply them not in the sense that “everything looks like a nail” when all you have is a hammer, but as a precise set of techniques to solve specific problems. Muzak, in its final days, was nothing more than a pipe dream. What people wanted was music — and choice. Those choices won’t come automatically. We may well have to hack them.

PETER KIRN is an audiovisual artist, composer/musician, technologist, and journalist. He is the editor of CDM and co-creator of the open source MeeBlip hardware synthesizer (meeblip.com). For six consecutive years, he has directed the MusicMaker’s Hacklab at CTM Festival, most recently together with new media artist Ioann Maria.

http://ctm-festival.de/

The post Minds, machines, and centralization: AI and music appeared first on CDM Create Digital Music.

NUGEN Audio updates MasterCheck Pro loudness, dynamics & codec toolset to v1.5

NUGEN Audio MasterCheck Pro 1.4NUGEN Audio has released version 1.5 of its MasterCheck Pro loudness, dynamics, and codec toolset. The update comes with enhancements embracing support for the Apple AAC (Advanced Audio Coding) iTunes Plus native OS X codec and also the PSR (Peak to Short-Term Loudness Ratio) measurement update to AES Engineering Brief 373. “I have to deliver […]

djay Pro 2 brings algorithms and machine learning to DJing

A.I.D.J.? The next-generation djay Pro 2 for Mac adds mixing and recommendations powered by machine learning – and more human-powered features, too.

When Big Data meets the DJ

The biggest break from how we’ve normally thought about DJ software comes in the form of automatic mixing and selection tools. One is powered by machine learning working with DJ sets, and one from data collected from listening (Spotify).

Automix AI is a new mixing technology. And hold on to your hats, folks, if the “sync” button was unnerving to you, this goes further.

When we say “A.I.,” we’re really talking machine learning – that is, “training” algorithms on large sets of data. In this case, that data comes from existing DJ sets. (Algoriddim tells CDM that was drawn from a variety of DJs, mostly in hip-hop and electronic genres.) Those sets were analyzed according to various sonic features, and the automixing applies those to your music. So this isn’t just about mixing two different techno tracks with mechanical efficiency – it’s meant to go further across different tempos and genres.

It’s also more than matching tempo. Automix AI will identify where the transition occurs, decide how long the fade should be, and apply filters and EQ. So, if you’ve ever listened to existing Automix features and how clumsy they are with starting and stopping tracks, this takes a different approach. Algoriddim explains to CDM:

The core of this tech is finding good start and end regions for transition between two songs, while also respecting the corresponding sound energies and choosing an appropriate transition accordingly (e.g. most likely EQ or short filter transition if you have two high energy parts of the song for the transition)

Then there’s “Morph” – which Algoriddim argue opens up new ways of mixing:

This actually goes beyond what a regular DJ can do with two hands. Morph not only syncs the songs but seamlessly ramps the changed tempo of the inactive deck to its regular speed as the transition progresses. E.g. in the past if you had a hip-hop song at say 95 BPM and an electronic track at 130 BPM, syncing the two and making a transition would leave the new track in an awkwardly rate changed state (even with time-stretching enabled). So as the transition starts, both songs (in this example) would be playing at 130 BPM but as we are doing a simultaneous tempo “crossfade”, the hip-hop track ends up being back at 95 BPM at the end of the transition. This ensures the tracks always play at their regular tempo and these types of mixes sound very natural, allowing for seamless cross-genre transitions.”

Also impressive: while you might think this sort of technology would be licensed externally, the whiz kids over at Algoriddim did all of this on their own, in-house.

On the Spotify integration side, and also related to automating DJing tasks, “Match” technology recommends music based on BPM, key, and music style. Existing Spotify users will be familiar with some of this recommendation engine already. Where it could be good for producers is, this means there’s an avenue by which your music gets exposed by algorithms. And that in turn is potentially good news, if you’re a producer whose music isn’t always charting the top of a genre on Beatport.

These “autopilot” features are all under your control, too: you can choose which parameters are used, choose your own tracks, switch it off at will – as you like. Or you can sit back and let djay Pro run in the background while you’re doing something else, if you want to let the machine do the DJing while you cook dinner, for instance.

Pro features, for humans

Okay, so at this point, djay Pro 2 may sound a bit like this:

But one of the disruptive things about Algoriddim’s approach to DJ software is, it has simultaneously challenged rivals both among entry level and casual users and more advanced users at the same time.

So, here’s the more “Pro” sounding side of this. Some of these are features that are either missing or not implemented quite the way we’d like in industry leaders like Serato and Traktor.

A new audio engine with master AU plug-ins. A rewrite of the engine now allows high-res waveforms, post-fader effects, higher-quality filters, plus the ability to add Audio Unit plug-ins as master output effects.

Integrated libraries. iTunes, Spotify, and music in the file system / Finder are now all integrated and can be viewed side-by-side.

Integrated library views bring together everything on your local machine as well as Spotify.

Smart filters. Set up dynamic playlists sorted by BPM, key, date, genre, and other metadata. (Those columns are available in other tools, but here you get them dynamically, a bit like the ones in iTunes.)

Keyboard Shortcuts Editor. There’s a full editor for assigning individual features to custom shortcuts – which in turn can also map to custom hardware or the MacBook Pro Touch Bar.

CDJ and third-party hardware support. Whereas some other players make their own hardware or limit compatibility (or even require specific hardware just to launch, ahem), Algoriddim’s approach is more open. So they’re fully certified by Pioneer for CDJ compatibility, and they include 60 MIDI controllers in the box, and they have an extensive MIDI learn function.

More cueing and looping. Version 2 now has up to eight cue points and loops, with naming, per song. (I recently lauded Soda for adding this.) You can also now assign loop triggers to cue points.

Single deck mode for preparation. Okay, some (cough, again Serato) lock you into this view if you don’t have authorized hardware plugged in. But here, it’s designed specifically for the purpose of making set prep easier.

Accessibility. VoiceOver support makes djay Pro 2 work for vision-impaired users. We really need more commitment to this in the industry; it’s also been great to see this technology from Algoriddim showcased at Apple’s developer conference. If you’re using this (and hopefully CDM is working well with screen readers), do let us know.

New photo / still image support.

And it does photos

Back to less club/pro features, the other breakthrough for casual users, weddings, and commercial gigs is photo integration. Drag and drop photos or albums onto the visual decks, and the software will make beat-matched slide shows.

The photo decks also work with existing, fairly powerful VJ features, which includes external output, effects, and the like. You can also adjust beat sync.

Still image support builds on an existing video/VJ facility.

Plus a no-brainer price

The other thing that’s disruptive about djay Pro 2: price. It’s US$49.99, with an intro price of US$39.99, on the App Store.

You’ll need Spotify Premium for those features, of course, and macOS 10.11 or later is required.

https://www.algoriddim.com/

The post djay Pro 2 brings algorithms and machine learning to DJing appeared first on CDM Create Digital Music.

NUGEN Audio updates MasterCheck Pro with FLAC & Opus encoding

NUGEN Audio MasterCheck Pro 1.4NUGEN Audio has launched version 1.4 of its MasterCheck Pro plugin for optimizing your mixes for today’s music delivery services. The update adds FLAC (Free Lossless Audio Codec) and Opus encoding to the indispensable, award-winning loudness, dynamics, and codec toolset. The term codec is an acronym for coder/decoder. Music streaming services such as Apple Music® […]

SoundCloud tries to allay fears, but streaming needs a business model

SoundCloud’s CEO published a post saying SoundCloud is here to stay and uploads are safe. But it isn’t just SoundCloud’s business that’s troubled.

Okay, first – the one thing you shouldn’t worry about is music you’ve uploaded to SoundCloud. As I wrote at the end of last week, you should worry if you have media that’s important to you that’s located in any one place without backups, SoundCloud or otherwise. But while there have been plenty of signs SoundCloud’s business is seriously troubled, that doesn’t necessarily translate to any indication you’ll lose access to the service.

SoundCloud co-founder and CEO Alex Ljung was left scrambling in the wake of deep layoffs to assuage user fears. He took to the phones with at least one celebrity user, Chance the Rapper, who reported a “fruitful” call with the exec on Twitter Friday.

Also on Friday, Ljung posted a plea on the company’s blog:

The music you love on SoundCloud isn’t going away, the music you shared or uploaded isn’t going away, because SoundCloud is not going away. Not in 50 days, not in 80 days or anytime in the foreseeable future. Your music is safe.

SoundCloud is here to stay. [SoundCloud blog]

Alex also refers dismissively to “an insane amount of noise” about the company.

But let’s back up. SoundCloud’s CEO can’t just shrug off fear and uncertainty when the company’s own messaging, actions, and even financial filings are largely responsible. Whatever’s going on with SoundCloud’s business, the company has lost control of its image. It’s hard not to view this “noise” as partly SoundCloud’s fault.

Co-founders Alex and Eric are each articulate and passionate advocates of music sharing. But the company has for years failed to articulate its business model. It’s talked about subscription services like SoundCloud Go, without being clear about how it can compete with entrenched competitors, and talked about advertising without being clear about how it will attract advertisers or how those ads will be effectively delivered. It’s been evasive about details of revenue and profit. It’s allowed bad press to accumulate, like allowing lavish office photos to spread just as financial filings were adding to concerns about its future. It has often failed to go on the record with press outlets (not mine, major press), while small rumor blogs flooded the narrative with leaked (and often inaccurate) information.

To see how badly SoundCloud’s media relations are going, look to recent reports by the likes of Forbes or even TechCrunch. That’s TechCrunch, who just last year were so bullish on SoundCloud that they said the company should be worth more than Spotify

Flash-forward to last week, and TechCrunch are reporting leaked accounts from inside the company’s headquarters and
questioning whether the company will survive.

The best SoundCloud could do by way of correction or response in this place was to say that the fourth quarter begins in 80 days, not 50, and that they meant they had money through the end of that quarter (that is, the end of the year) – but that means we’re not any further along than when Ljung initially made that same statement in a financial statement in January. You can watch the messy back and forth here:
SoundCloud Responds to ‘Extensive Inaccuracies’ in Article Claiming It’s Almost Out of Money

[Indeed, TechCrunch has reason to complain here – SoundCloud doesn’t specify what it means by “extensive inaccuracies,” and actually appears to confirm some of the main gist of the article.]

Presumably these layoffs were planned for some time, so why did SoundCloud appear to be improvising its message to the press and its own staff?

And this problem isn’t a new one in summer. Way back in January, the apparent failure of revenue plans to keep pace with growing costs were fueling acquisition predictions. Now we have vague platitudes from the CEO that the company intends to remain independent, without any material on how they will do that. (That is, even after 40% staffing cuts, they’re still not talking about having money after the fourth quarter, unless by “foreseeable future” Ljung only means he can forsee 2017!) Here’s Fortune back at the end of the year; we actually know very little new information since then:
Here’s Why SoundCloud Will Likely Look to Be Acquired Soon [Fortune]

I know SoundCloud can do better, having covered the company since its 2008 founding. I know its founders can do a better job of messaging than this, too, having known them almost as long. Rather than simply imploring its users to help, they need to provide a better picture as soon as possible as to how revenue growth will work versus costs – particularly now, having cut some of the staff who were responsible for making that revenue growth happen.

A Spotify timeline. Photo (CC-BY) Jon Åslund.

Not only SoundCloud

That said, I think SoundCloud are unfairly bearing the brunt of bad press and angry musicians.

Let’s not mince words: right now, the whole model of streaming appears economically broken, and surely all the major players deserve some share of the blame.

Talk about a rock and a hard place – maybe “buried under a pile of rubble” is more apt.

Content creators and owners believe they should get paid for music being streamed. So you’ve got the industry that represents them asking for higher royalty rates.

The problem is where the revenue to pay those royalties is coming from. Listeners don’t appear to want to pay much for subscription fees. That’s at least partly why Spotify and SoundCloud and others aren’t showing profitable results. Even if you don’t buy their arguments (lavish offices and huge headcount being evidence), there’s still a fundamental problem here. If users pay a flat fee for a subscription, then the company loses money the more they listen to the service – because royalty costs accrue. SoundCloud here actually has an edge, in that not all of the music uploaded requires a license – think spoken word and unreleased music. But SoundCloud hasn’t yet proven that they can make this work, either. (We’ll see if those staff cuts or other budget trimming helps.)

Advertising is the one thing that will grow with increased listening, at least in theory – more listening means more revenue for ads. But listeners and even content creators have been resistant to advertising. And selling ads in sufficient volume and with significant value means you need to have a talented staff able to liaise with big agencies and advertisers. Google is the one tech company who seem to have built a significant competency in the ad business, but they claim they’re not making money on ads, either.

And it gets worse. Largely missed in all the coverage of SoundCloud last week (but observed by some CDM readers), it’s really YouTube that dominates streaming. The Washington Post has just painted a bleak picture of the value of those YouTube plays to music.

Why musicians are so angry at the world’s most popular music streaming service

In a pot-calls-kettle-black argument, YouTube weirdly warns of the dangers of consolidation in big players:

“The industry should be really, really careful because they could close their eyes and wake up with their revenue really concentrated in two, three sources,” said Lyor Cohen, YouTube’s global head of music, referring to Spotify, Apple Music and Amazon Prime Music.

Right, so it’s better if it’s concentrated in four, and the fourth is Google? Huh?

The real danger here seems to be a race to the bottom. Apple, Amazon, and Google can all afford to lose money on streaming, turning it into loss-leader business for other revenue streams. SoundCloud, Spotify, and other tech companies can afford to lose money by repeatedly turning to outside investment. (It’s absurd that we’re still calling this “runway” with those companies, as the business is now around a decade old at least. The runway metaphor only works if you take off at some point. A “hole in the ground into which you throw money” metaphor is what we seem to have here.)

I wouldn’t normally compliment the record industry, but to the credit of groups like the RIAA, at least they’re exerting some pressure up. The problem is, even a $7 royalty per 1000 streams may prove negligible to smaller artists and labels – and if the business that pays that royalty can’t survive, it’s a moot point anyway.

So, uh, how’s everyone feeling? Super… happy? No?

Of course, the buzzword that everyone seems to be running to at the moment is the blockchain – offering decentralized content and paying creators more directly. But describing one part of a larger solution isn’t the same as describing the whole solution. Will listeners embrace micropayments for music, or will they find it a hassle? What will make them migrate from services they’re already accustomed to using – and in which they’ve already assembled playlists and preferences? What about the fact that services like Apple’s are already integrated with the listening devices they own? How do you convince listeners to change their mind about what music should cost, when they’ve already grown accustomed to $10 monthly fees – or, very often, no fee whatsoever?

It isn’t all bad news. People are listening to more music. Streaming isn’t a nonexistent business – it’s US$7.7 billion in the United States alone. Someone, somewhere is actually earning money.

Also, because of the cost of PR and building fanbases, and the potential revenue earned from paying live (or selling physical goods), a lot of musicians I’ve talked to really do appreciate the promotional value of online streams. There are plenty of cases where giving away streaming music is viable – because you might then sell people vinyl, for instance.

And, look, while all of this shakes out, musicians and labels continue to pursue a strategy that caters to building relations on all these services. Some of them have great success stories with YouTube, with SoundCloud, with Spotify.

But maybe that’s the point. It seems to be the businesses in between that are non-functioning – or (in the case of futuristic blockchain propositions) just not ready for primetime.

Musicians and labels keep doing the hard work of making the music and fighting to get it heard. Yet investment and attention pours into the middleware between us and listeners – and that middleware really isn’t working terribly well.

At the very least, it seems totally valid to me that people who make music have reason to be frustrated. I think we should continue talking about our own solutions. And I’d like to see the captains of industry – music industry and tech industry alike – take some greater responsibility for what’s gone wrong and how it might go better. Well… one can dream, anyway.

So, uh… vinyl? Cassette tapes? Eight tracks?

Erm… happy Monday?

The post SoundCloud tries to allay fears, but streaming needs a business model appeared first on CDM Create Digital Music.

SoundCloud cuts 41% of staff as streaming music business melts down

There’s a famous line in business by Tom Peters, in his book The Circle of Innovation: “you can’t shrink your way to greatness.”

But shrinking is exactly what SoundCloud now has to do, with its survival – let alone any ongoing greatness – at stake. As founder Alex Ljung puts it in a blog post for the company today, massive headcount reduction and the closure of two offices is necessary to put the company “on our path to profitability and in control of SoundCloud’s independent future.” The implication is, without a buyer, the company may not last without cutting staff.

Asked for comment, SoundCloud pointed CDM to that post:

A note from Alex Ljung

SoundCloud will lose a lot of the people who made the service valuable. 173 out of 420 employees – 41% of staff – are being made redundant. San Francisco and London offices are closing, leaving New York and the headquarters here in Berlin. (That may have implications for Berlin’s reputation as a European Internet capital, as well, as SoundCloud has been its best known poster child.)

I know some of these people personally. I’ve seen what they bring to the service and our music community in general. I’ve also seen how significant SoundCloud has been in helping musicians share music and people to discover that music, its impact on record labels, on artists getting bookings … on daily life.

I think artists and ex-employees alike could feel legitimately betrayed by the course music streaming has taken. SoundCloud at least is increasing revenue. Ljung says the company has “more than doubled” revenue in the past 12 months, without citing specific breakdown of producer subscriptions, listener subscriptions, and advertising. But the issue is how revenue compares to costs.

Now, ironically, the writing has been on the wall for a decade. Ten years ago – and one year before SoundCloud was founded – Pandora co-founder and ex-CEO told CDM he thought streaming rates would shutter companies. The weird part of this is, he may have been right – it’s just that an ongoing influx of investment has prolonged that failure over the years.

If Streaming Rates Stand, “We’ll Have to Shutter”, Says Pandora Founder

If it seems greedy that he’d suggest such a thing, one reason is that there aren’t such royalties collected on radio broadcasts.

Whether you want to blame the services, tech giants like Apple and Amazon, or the music industry for setting rates, the business model just doesn’t seem to add up anywhere. And 2017 could be the “s*** hits the fan” moment as it becomes ever clearer that no one is able to turn that business model into a win.

Just last week, co-founder and returning CEO Tim Westergren left Pandora. That company has never made a profit, and it seems new investors Sirius XM (satellite radio company) have other plans.

Then there’s Spotify. As its revenues and number of users grow rapidly, its losses are actually growing even more rapidly. That should mean that Ljung’s comment about growing revenue is as much a red flag as it is encouragement.

Spotify’s Loss More Than Doubles Even as User Growth Surges [Bloomberg]

Noticing a trend here? Pretty much anyone in the streaming business is losing money. That overall picture also will rule out some acquisitions, or reduce the price. And it’s not surprising that this combination might frighten away some investors.

CDM readers and associates frequently compare Bandcamp to SoundCloud. But perhaps if any comparison is apt, it’s because of the contrast in business models, growth rate, and intended audience. Bandcamp remains a niche site for people to consume music, not only as free streams, but as downloads, physical media, and in the form of merchandise. It’s the always-on, “tap water”-style streaming that is having trouble.

To state the painfully obvious, it’s also troubling to look at the streaming players who are thriving. Facebook has stayed out of music (unlike Russian social media network VKontakte). But three other big tech giants – Amazon, Apple, and Google – are able to offer streaming services as “loss-leader” offerings, directing sales elsewhere. Apple may lag Spotify, with 27 million users to Spotify’s 50 million. But then Cupertino doesn’t need Apple Music to turn a profit, since the company can instead sell iPhones, iPads, and Macs.

It’s just as easy to find music on YouTube – which also spells further pain for artists and labels.

Music press have been quick to jump on SoundCloud, often without much to back them up. But now, I believe it’s reasonable to sound some alarms. Staff cuts this significant could slow growth and curb the efforts that would expand revenue. They suggest serious financial obstacles. And there’s still not a clear picture of how streaming will be sustainable as a business model – not for SoundCloud, and not for the entire industry.

And the implications there go far beyond SoundCloud’s offices. They should raise serious questions about what a record label is, how it collects revenue in the digital age, and how much control artists and publishers will have on their music being shared and discovered.

Of course, that absolutely means now is the time to talk about alternatives, including innovative solutions like Blockchain-powered sharing and the like. But the popularity of SoundCloud and Spotify for finding and playing music is going to be a tough benchmark to match.

Whatever happens next, it’s going to involve some major changes. And if these companies do start to contract, a lot of the talent that was working on the problem is going to wind up elsewhere.

The post SoundCloud cuts 41% of staff as streaming music business melts down appeared first on CDM Create Digital Music.

Producers and DJs can now sign up to get paid for SoundCloud plays

SoundCloud’s ambitious goals for being the place where people share and discover music has always left it with a challenge. On one hand, it has to keep encouraging you to upload music – your tunes, your remixes, your DJ sets. It can’t just be a site for major label content, because then it loses to Apple and Spotify by default. On the other, it has to satisfy the needs of right holders – including when you upload music that they own. That’s an issue with your DJ set, of course, but it could eve be an issue with your own music, if you’re signed to a label.

Musicians and DJs have often assumed that these two interests conflict, but that’s not necessarily so.

First, the more SoundCloud attracts popular, major label content, the more likely it is that people listening to that content will find your music, too. That’s true in theory, anyway – and SoundCloud’s discovery algorithms are specifically tuned to introduce other music. SoundCloud has popularized a breakthrough track here or there, and there’s even word via Chance the Rapper that streamed music (including SoudCloud) could be up for Grammy awards. So this is really a thing.

Second, SoundCloud’s legal battles have actually made it easier for DJs and producers to upload their music without annoying takedowns.

But the ultimate prize is, SoundCloud will pay you for streams. That on its own might not be relevant, but SoundCloud gives individual artists far more control than Apple and Spotify. Only SoundCloud lets you upload content directly, at will, and the statistics and social controls available far exceed anything else. So combined with the ability to get paid, even a little, this is a big deal.

Obviously, the way this works is for SoundCloud to use its paid options (and presumably advertising revenue, though we’ve not seen a lot of that) in order to direct funds to the artists.

The details are a bit sketchy, and the rollout isn’t instantaneous. (That’s unfortunately been a pattern with SoundCloud, though I don’t know to what extent to blame the Berlin company and to what extent to fault the byzantine entities of the record industry.)

Basically, what SoundCloud are announcing is that the Premiere offering will soon be available to DJs and producers. For now, you can sign up; there’s no word yet on who will get included when.

Maybe the most interesting detail here is actually that SoundCloud specifically calls out DJ mixes (that’s the first time they’ve done that on the record, officially), and remixes. The sets are a big deal, because licensing them has been a hurdle in the past. And while Mixcloud ostensibly offers better tools and more solid licensing, I’ve never once talked to a DJ who was as happy with the plays and engagement they got on Mixcloud as they were with SoundCloud.

Here’s the signup:
https://on.soundcloud.com/premier

The news comes at an interesting time. It seems increasingly likely that SoundCloud could soon face heated-up competition, especially as DJ mixes spread. Apple took some (small) steps into that space as far as licensing, and more recently, Spotify purchased technology that lets it better identify songs and stems.

The acquisition alone is bad news for SoundCloud – even if Spotify doesn’t use the tech, they just blocked SoundCloud from getting to it. (And in acquisition-happy tech, sometimes even that’s the goal.) But it could also be the basis of Spotify adding features that compete with SoundCloud. I doubt seriously that you’d get a Spotify/SoundCloud mash-up, but it could cover something like DJ tech.

I’m surprised, though, that no one has speculated that a Spotify acquisition of SoundCloud could go hand in hand with buying Sonalytics, since the latter would shore up the legal basis of the former.

Meanhwile, there are the usual swirling discussions about SoundCloud running out of money, which we’ve heard now for years on a regular basis. Since the company was founded, it has routinely had to go ask investors for more cash to extend runway – to the point of maybe “runway” isn’t really the word for it any more.

In other words, running out of money isn’t news. It’s a question of whether investors want to keep investing. But I wouldn’t trust anonymous sources for reliable information there, which is what some more speculative sites of relied upon.

Getting bought would be another route, and all of this could simply have to do with negotiations and asking price. Some of this uncertainty I think isn’t just a reflection of SoundCloud and their business model, but the uncertainties of streaming music in general. Because performing rights organizations and publishers and labels effectively set their share of money, and consumers have set their expectations of what they want to pay so low, businesses get badly squeezed in the middle.

Anyway, it’s a good thing we’re musicians. We have none of that uncertainty – we know most of what we do won’t make any money.

But I’m totally biased. I’m rooting for SoundCloud because they provide tools for musicians that no one else does, for the moment. And this move makes the stuff they make for us more useful. And for all SoundCloud’s flaws, right now, we need all the tools and control we can find.

DJs and Producers now invited to SoundCloud Premier

I will see if we can get SoundCloud management to go on record about any of this; stay tuned.

The post Producers and DJs can now sign up to get paid for SoundCloud plays appeared first on CDM Create Digital Music.

Spotifyプレイリストの自動作成可能、デジタル時代のアナログターンテーブル

9a0e54db5fbb05f6d3aee393c9e5a027_original

 

イギリス・ロンドンのマスタリングスタジオ+レコードレーベルGearbox Recordはハイクオリティなサウンドとデジタル技術を合体したGearbox Automaticを発表しました。現在クラウドファンディングKickstarterにて出資者を募っています。

ここ数年間、急激に人気が復活しているアナログレコード。そのアナログレコード人気に後押しされ、現在様々なアナログターンテーブルの新機種がリリースされていますが、例えば昨年劇的な復活を遂げたTechnics は高級オーディオマニア向けターンテーブルSL-1200Gをリリース(およそ350.000円)。一方ではデジタルフォーマットへの変換が可能だったり、 アンプやスピーカーを内蔵していたり、手軽にレコードを聴く環境としての格安なターンテーブルも多く見かけます。

Gearbox Automaticはこの2つの溝を埋めるターンテーブルです。大きな特徴は真空管フォノステージ搭載していること。この段階でかなり高いサウンドクオリティが約束されます。また一方では、BluetoothやWifiとの接続も可能で、アナログサウンドをデジタルに変換することやBluetoothスピーカーからの出力も可能となり、今の時代のステレオ環境との相性もぴったりです。面白いのはプレイしたレコードが自動的に認識され、Spotifyのプレイリストに加えられることです。価格が£399(およそ57.300円)というのもかなり魅力的です。詳しくはkickstarterよりどうぞ。

bcb1765f2ceeb4fbb7388e96adec6193_original