An update, future of the DAW, in defence of the RIAA and ADE
The AI voice modelling debate is heating up…
Hello!
First of all apologies for the extremely long delay between this and my last newsletter.
It was a series of unfortunate events, starting with a string of cancellations of interviews I had planned towards the end of August (the music industry and work in August do not mix, it seems), followed by a holiday in Ibiza which ended up getting extended until mid-Sept and then, I got extremely sick! It wasn’t COVID, but it went on forever, about 11 days of feeling like shit in the end. And that meant I was waaaaay behind on other things, and everything got pushed back, including some interviews that had originally been rescheduled from August! My fault this time.
So I’ve since spent between the start of October and now catching up on a whole load of work, including more writing and prepping for a conference called Unwrap, which took place last week in Kortrijk, Belgium, and then ADE. So I’m sorry for letting Future Filter fall by the wayside.
And during all that, a piece about the future of the DAW I wrote for DJ Mag finally came out. I hinted at a while back in my interview with Lex Dromgoole from Bronze. You can read the Lex interview here, and you should, as his thoughts are super interesting and always welcome.
Thank you again for subscribing and to all I met at Unwrap – hope you enjoyed it, definitely a conference to watch for the future. See you in Amsterdam!
Future of the DAW
The full Future of the DAW piece ended up triggering a whole bunch of conversations both privately and publicly. I do find the sense of loyalty people have towards DAWs pretty funny. It’s kind of like the old Mac v PC days, although I think people are more multi-DAW’d now than ever before, partly cause they’re a lot cheaper than they used to be, so it’s more feasible to jump around.
What was even more interesting is that we had to push back publication three times as more and more things started to get announced – namely, FL Cloud coming to FL Studio, Bitwig and PreSonus’s new file format that supports both DAWs (and aims to support more) and Pro Tools announcing their Sketch app. After years of stagnation things were starting to happen just as we were pointing out how badly things needed to happen, which was nice as it felt like we were on the right track.
One of the main takeaways people came back to me with was that it’s not the DAW’s responsibility to make the user creative. And BOY HOWDY do I agree with that. I really do. Going out of your way to ‘aid’ creativity can do more harm than good, as it often sends every user down the same creative path. But there’s a difference between actively encouraging creativity and actively blocking it. I would say, personally, that the DAW does more to block creativity than it does to aid it, as it stands. Not every DAW, not every feature, but generally across the board that’s how I feel.
Creativity is not static, it is a malleable and forever-evolving concept whose goalposts are constantly being adjusted and reevaluated. At the same time, creative tools shouldn’t dictate creative outcomes. But there is a balance, and that balance is currently off-axis. I was happy to see FL-Studio’s new cloud features as it feels like the beginning of a new journey in what the DAW can be.
For the piece, I spoke to Meng Ru Kuok from Bandlab, David Ronan from RoEx, Joshua Hodge from the Audio Programmer community, Tim Exile from Endlesss, pro audio consultant Scott Simon and Yotam Mann from Sounds.Studio. The TL;DR of the piece is basically: the DAW is really dumb and siloed, but still super powerful and flexible, the future of music making is social, and they need to at the very least make DAWs cloud-connected so they can be properly collaborative, while also adopting new tech APIs like generative AI tools that are also cloud-based. Thoughts?
It was well over ten hours of interviews, some of which I barely was able to use, which was a shame. I had a whole section on how skeuomorphism and the ongoing fetishisation of analogue is one of the reasons we are stuck in a creative rut when it comes to new creative music tools, but it had to be cut out. Feel free to buy me a Jupiler at ADE to hear more about that.
ADE Incoming
Speaking of which, the big week is upon us. I haven’t been to ADE since 2019, which was my 11th ADE in a row. It’s by far my favourite conference, I think because it’s in October, and it’s not really a sunny city (and is usually pissing rain, with added ice wind) people tend to go in more with a business mindset and want to achieve things and have good meetings before bedding in for the quieter winter months, as opposed to IMS and WMC where the sunny weather can be… distracting. I’m there from Wednesday morning – when I’ll be speaking at ADE Labs – until Saturday, and I don’t think I’ve ever had so many meetings. BUT, if you do want to have a beer or a coffee do reach out and hopefully, we can squeeze it in, even if it’s just a 10-minute catch-up. Also please do let me know if you’re hosting any events or mixers. I intend to also do a bunch of interviews at ADE to get back on track for Future Filter and start sending out these more regularly again.
Regulating a Revolution
Full disclosure: I am working very closely with a platform called Voice-Swap and there’ll be a lot more news on that post-ADE, but wanted to make that clear as we operate in the space I am about to comment on. If you think my pro-artist approach is a result of my bias due to Voice-Swap, that’s fine (it’s not, as my newsletter in July has the same ethos), but it’s important to me to always remain transparent about things I may have a stake in that relate to what I’m saying on Future Filter.
Since we last spoke, there’s been a whole bunch of developments in AI – shocking. A lot of the time when we discuss AI we’re talking about an imagined future, where we are either assuming the legislation has been sorted out and we are free to let our creativity run wild, or we enter a copyright-agnostic Wild West where anything goes. The past few weeks have shown some real, tangible signs of where things might go next and it’s largely being led by the RIAA.
A few weeks ago, the RIAA forced Discord to close one of its servers – AI Hub – that was being used to train and share unauthorised voice models of famous singers. The RIAA’s initial request dates back to June this year, but it was only in October the server officially went dark.
Next, the RIAA submitted a request to the US Trade Representative (USTR) to add “the category of AI voice cloning in its annual list of entities that reportedly promote piracy or counterfeiting”. Essentially adding unauthorised voice cloning services to the same list you’d find The Pirate Bay, TorrentFreak and other illegal file-sharing platforms. RIAA went as far as to name one platform that was responsible for these models – although there are many – Voicify.ai, created by a 20-year-old student according to Music Business Worldwide.
On top of the news from RIAA, Google also stopped RVC training on Colab for its free users – basically restricting access to custom voice model training to pro users only. This coincided with their partnership with UMG, which we discussed before. Maybe it’s a coincidence?
Not everyone was happy about the news – Daouda Leonard, Grimes’ manager, who’s obviously been at the forefront of the AI voice model story so far, X’d out: “Always sad to see new cultural communities form and get crushed by bureaucracy.”
History Repeating Itself?
Almost everyone can agree that the heavy-handed approach the record industry took – often led by the RIAA – in the early days of piracy was OTT. That heavyhandedness played a large part in stopping the industry from reacting quickly enough to a new era of music consumption, which music fans had already adopted with or without permission. Even the iTunes Store, in hindsight, felt like a stopgap, and it was only really streaming that ‘saved’ the industry, almost 15 years later, extra large caveats and elephants in the room notwithstanding.
This time though, I do agree with the RIAA’s approach, to an extent. Piracy and the P2P sharing of files feel like something very different to the cloning of someone’s voice. To me, it’s a lot more invasive to take a voice, which is one of the most personal and intimate ways in which humans identify each other, and use it in your own work to say or sing something that person never sang without permission.
A music file is a static piece of audio, whereas an AI model is the opposite. It’s full of variables by design – even if you run the same acapella through the same model twice, it won’t be the same output, or at least won’t be identical. Everything we know about protecting IP online refers to static files that are generally traceable and identifiable. Generative AI upends that presumption and therefore requires new solutions. We’re dealing with something completely different when compared to an MP3, and therefore we can’t simply dismiss the RIAA’s approach as repeating the same mistakes of the early-00s.
Is it predictable that this approach would be taken? Yes. But is it the wrong path? I don’t think so.
There is no doubt in my mind that voice modelling will become both a bona fide creative tool and a legitimate practical tool for countless industries – in fact, I’m working with DJ Fresh’s Voice-Swap for that reason – but none of that can happen until we lock down protections for those who want to experiment with this tech.
Suspect Sentiment
As part of the talk I did at Unwrap conference last week, I showed two different slides on AI sentiment, from two different surveys, both asking ‘artists’ questions about their thoughts on AI. One was from Ditto Music, and the results have been widely circulated already. The other was PRS for Music, the UK PRO. Putting the results side by side shows how far apart even artists are in their sentiments towards AI.
Some insights:
Where 59.5% of artists surveyed by Ditto Music said they “already use” AI in their music projects and 47% said they would use AI for their songwriting, 71% of the PRS respondents said that they were not currently using AI for music-related activities, while 55% said that they would be open to using AI for music-related activities in the future. The overwhelming PRS majority, however, agree that transparency and remuneration are vital for AI to be a positive force in music, with 93% believing creators deserve to be compensated if their music is used for AI-generated content and 89% feeling AI tools should be transparent about how they generate AI works.
Of course, the demographic of a Ditto Music poll and a PRS one are wildly different, one being a distributor for independent, likely younger artists, while PRS is a legacy industry body with paid-up members who tend to make their living solely from their art.
The point remains: sentiment, even among working or semi-pro artists still remains mixed. And yes, there are always tech bros who want to steamroll ahead without concern for copyright or IP, as Marc Andreeson’s bizarre “techno-optimist” essay from just yesterday outlines, pining two sides against each other isn’t going to move us forward to an equally palatable plateau. It’s the default state of the modern internet to pick a side and bed in, but in this case (in fact, in every case) it’s not helpful or productive.
The creative power of this tech is so strong, that it WILL happen, and it WILL become a standard part of the music-making toolkit. Current sentiment implies there’s still a long way to go, but respecting artists' and rightsholders’ concerns shouldn’t be seen as an inconvenience, despite the mistakes of the past.
Other recommended stuff from the internet
I just finished Meghan O’Gieblyn’s excellent ‘God, Human, Animal, Machine’ and have started Cliff Kuang’s ‘User Friendly: How the Hidden Rules of Design are Changing the Way We Live, Work & Play’, which is really interesting so far.
Eleven Labs just introduced a new dubbing technology that really blew my mind. You can paste a YouTube link and select a language and in less than five minutes you get the video back, dubbed in that language. Try it.
Shawn Reynaldo made his recent newsletter about ‘Bandcamp anxiety’ free to read after the sad news that 50% of their staff were being made redundant following the sale from Epic Games to Songtradr. It’s not yet clear the final details and how the recently formed Bandcamp Union will play into the redundancies.