Automating the Mix: How RoEx Helps Musicians Get Their Music Out There
Automating the Mix: How RoEx Helps Musicians Get Their Music Out There
David Ronan on AI, Creativity, and How RoEx Bridges the Gap Between Raw Creativity and Polished Tracks.
In the vibrant world of music production, the journey from creative spark to finished track can be fraught with technical hurdles. Anyone who has spent a lot of time with creative types will probably agree that the biggest obstacle to putting their work out is the idea that "it's not ready". In the world of music, this often relates to the mixing and mastering of the track. Although lots of producers and musicians also mix and master their music, many don't, which can become a blocker to putting out music. That is why when I first heard about RoEx, I got excited.
RoEx is a London-based mixing and mastering startup on a mission to help creators and musicians create outstanding sound quality that resonates with their audiences. A spin-out from Queen Mary University of London, the company builds on pioneering audio mixing research to develop advanced AI solutions that unlock creative potential, making audio production expertise more accessible and boundless than ever before.
Its two consumer products are Automix, which allows musicians of all skill levels to produce professional-grade audio mixes and masters, and Mix Check Studio, a free tool that has provided actionable feedback on over 500,000 mixes and masters to date.
RoEx’s Intelligent Audio Engine, Tonn, allows B2B partners to integrate start-of-the-art mixing and mastering features with their existing software, using RoEx's technology to unlock quality at scale.
To date, RoEx has been awarded a total of £300,000 from Innovate UK, the UK government’s innovation agency, including a recent grant to fund the development of its pioneering style transfer tool, ProStyle.
In this interview, we explore the journey of David Ronan, from his beginnings as a music technology student to founding RoEx and leading innovations in AI for music production. Ronan shares the experiences and challenges that shaped his path, offering insights into the changing world of music technology and entrepreneurship.
Thanks for taking the time out to talk today, let's start off with a bit of background about yourself. What made you want to get into the Music AI space. What experiences led you to founding RoEx?
I was really into electronic music growing up—big fan of Warp Records, Aphex Twin, Squarepusher, and artists like that. I eventually started making and producing my own music when I was in university. I wasn't brilliant at it, but I decided I wanted to study it more formally.
After finishing my undergraduate degree, I pursued a master’s in music technology. I completed a year of that and absolutely loved it. It was a quick transition for me. Then I moved to London for broader opportunities, worked for a few years as an engineer, but I knew I wanted to return to music.
I ended up doing a PhD at Queen Mary University of London, specifically at the Centre for Digital Music, which is one of the biggest research groups in the world for music tech. My supervisor was one of the founders of LANDR so my research focused on intelligent audio production tools.
I worked on that for four years, with a brief interruption to work for Native Instruments in their Berlin office. Working as part of their R&D team was a lot of fun and marked the beginning of my R&D background, along with my PhD work.
At the end of my PhD, I had some intellectual property related to automated mixing. However, it was limited—it only ran in a lab, took a day to produce results, could only mix four tracks, and was in mono, which was essentially useless. But the science behind it was sound.
Without any money or startup experience, I decided it wasn't the right time to commercialise my work. So, I went to work for a company in London called AI Music.
You might guess what they do. Essentially, we used AI to take existing samples and recordings and create new music from them. It was licence-free music for advertising and generative music before generative AI took off.
I was the head of research there for four and a half years. When AI Music was acquired by Apple, I had the opportunity to join Apple, but I chose to start RoEx instead.
Taking that leap of faith was a big decision, especially since my daughter had just turned one. I started RoEx, spun out from the university, and secured our first round of investment.
Now we’ve built a big team, released two products, and processed over half a million tracks, and things have really taken off.
A bit about our platform: Our flagship tool, Automix, is designed so that musicians can upload their recordings. Ideally, it's for those who are very creative but either don't want to handle the mixing themselves or can’t afford to hire a mix engineer, as they can be expensive.
With Automix, users provide some basic mix settings and can do a bit of cleanup if needed—since sometimes recordings come from less-than-ideal conditions. We then mix and master the track for them, so it's ready to go on Spotify or any other platform in minutes, instead of days, and at a very low cost.
In addition to Automix, we have another tool called Mix Check Studio. This free tool lets you drag and drop a mix and master track you’ve already completed. It analyses the track and provides feedback on how to fix it in your DAW.
We’re also about to release the next iteration of Mix Check Studio. I can’t say too much because it’s top secret, but it’s coming very soon.
The idea for this all started when I was studying for my master’s and learning how to mix and master. I remember the first time I opened a DAW—I was overwhelmed by the sheer number of buttons and options. What’s MIDI? How do I use a compressor?
It took a long time to understand it all. I realised that to excel, you need to invest time in learning. Mixing involves two parts: the creative side and the technical side, such as balancing and setting EQs for different instruments. There was nothing particularly creative about that.
So I thought, surely there must be a way to automate this process. That was 14 years ago, and now we’ve automated it. That’s the origin story of RoEx.
What were some of the biggest challenges you faced during the development of RoEx, and how did you overcome them?
I guess the toughest part was, as I mentioned, the prototype that ran in the lab. It would take forever to get a result, so it required a lot of development time.
Luckily, I have experience as a software engineer, which allowed me to dive in, refine, develop, and optimise it. I managed to get it running on the cloud, which was a huge challenge.
There was also the constant doubt: Am I doing this correctly? Can I trust my own ears? Does it actually sound good?
The turning point came when I was able to get a prototype—or rather, a beta—ready. I launched the beta in September 2022, and it was only when people started using it that I began to learn how to refine and improve it. Feedback from real users helped me understand how it sounds and how to make it better.
I often hear mixers talk about the importance of 'feeling' in their work. While the technical aspects of mixing can be handled by automated processes, what’s your thoughts and approach to incorporating the human touch that many believe is essential and often sought after in AI models?
Initially, when we launched the beta, it was a completely hands-off process. You would just get the mix back and either like it or not—that was it.
We found that this approach didn’t resonate very well with people. So, we’ve since adjusted our method. Now, you receive a preview that allows you to tweak various elements. Currently, you can adjust the levels, and soon we’ll be adding options to tweak the panning, EQ, and compression. The idea is to make it a suggestive tool rather than giving you a final mix that you just accept or reject.
This way, you’re part of the process and maintain some creative control. At the end of the day, it's your art and your music. You don’t want to be told exactly how it should sound; you want to make it sound how you envision it. While mixing and mastering can be technical and challenging—many people aren’t sure what changes will sound best—our goal is to get you 95 to 99% of the way there. Then, you can make the final tweaks that make it truly yours. That way, everyone benefits.
Are there any accepted "truths" about AI's role in music creation that you disagree with based on your discussions with musicians? What's your contrasting viewpoint and evidence?
I guess everyone's worried about generative AI—like Suno and Udio, which have released models in the last year that create really good-sounding generative music. People are concerned that it might replace music as we know it, but I genuinely don’t think that will happen. You’re just typing in a few words and getting a song back. While there’s instant gratification and an initial "wow, that sounds amazing," it’s not music that you’ve put your energy, life, and emotions into. It feels very throwaway.
Creating art is all about the process, and that’s really important. I don’t think AI will replace that. For example, people go to see a band because they want to experience the band’s performance and the lore surrounding it. AI doesn’t offer that same experience.
Of course, AI will serve its purpose—it’s going to be useful for certain tasks, though it’s not without issues. For instance, AI training on copyrighted material without compensating artists is a problem. Every artist should be compensated for their work, and that's a separate topic.
In terms of replacement, I don’t think AI will replace human creativity. Instead, it will act as an assistive tool. Much like what we’re doing with our technology, AI can help with challenging tasks like mixing. If we view AI tools through this lens, it can allow us to focus more on the creative aspects of our work. I recently saw a brilliant quote that sums it up well: when I think of AI, I imagine it doing tasks like washing my clothes, not the creative hobbies I enjoy. That’s exactly what we aim to achieve—using AI to handle the routine stuff so we can focus on the fun and creative parts.
How do you see the future of AI in the music industry evolving in the next 5-10 years? Where do you envision RoEx in that future?
With the advent of generative AI, legal frameworks are going to need to change and catch up. For example, companies like Suno and Udio are currently under scrutiny and facing lawsuits from the RIAA. I think we’ll eventually see the development of licensing models to ensure that artists are compensated. Technology will continue to evolve, but laws often lag behind, so there will be a lot of adjustments needed.
There’s a lot of anger about these issues right now, and I hope that things will come to a head soon, leading to fair compensation for artists. Despite these challenges, I see a huge increase in content generation in the coming years. AI tools will assist in generating melodies, drum beats, and other elements, which can be incredibly useful. For example, I was always bad at programming drums, so having AI assist with that would be a huge help.
It’s exciting how rapidly things are moving. There’s something new every week, and it’s hard to keep up. I’m really excited about the future. I think DAWs (Digital Audio Workstations) will increasingly move towards the cloud and integrate more AI features. Many DAWs have been around for over 20 years, and they’re starting to incorporate AI tools, like Logic's source separation and automatic mastering features, which is fantastic.
For us, we’re currently working on DAW integrations. Our goal is to be deeply integrated into the creation process, helping people sound their best despite any technical or financial hurdles. We aim to continue evolving our tools to fit seamlessly into DAWs, making the creative process easier and more accessible.
What advice would you give to aspiring entrepreneurs looking to enter the AI or music tech space?
First and foremost, surround yourself with good people. Good people attract other good people. This is a self-fulfilling prophecy—if you’re a good person yourself, you’ll naturally attract others who are good. Lean on friends and family, especially in the beginning, because starting out can be tough.
There's a quote often attributed to Elon Musk, even though I'm not personally a fan of his, that resonates: when you're starting out, it can feel like you're chewing glass and screaming into the abyss. You’re small, no one knows you, and it takes time. Persistence is key. Keep posting on social media, engaging with people, and building your network. It won't be an instant hit—though some might get lucky—most success comes from hard work and persistence.
Talking to people who have been through the entrepreneurial journey is invaluable. They often want to pay it forward, having relied on others themselves at some point. In my experience, people are usually eager to help someone tackle a new and challenging endeavour.
Lastly, be genuine. Authenticity is crucial; people can usually sense insincerity from a mile away. Being yourself and being honest will serve you well in the long run.
Is there anything else you'd like to share about RoEx or your personal journey that we haven’t covered?
I’m incredibly grateful for all the people I've met along this journey. Many of them have steered me in the right direction, and I truly believe that surrounding yourself with good people leads to good things. Looking ahead, I’m really excited about the coming year for us. We have some fantastic new products and extensions of our current offerings on the way.
One of the most rewarding aspects of this job is seeing people enjoy and benefit from the tools we’ve built. Receiving feedback like, “Wow, you changed my life; I can finally get my music out there,” is incredibly fulfilling. For entrepreneurs, that kind of impact should be a guiding principle.
Our goal is to empower people by handling the technical aspects musicians might worry about, giving them the confidence to focus on their creativity. We take care of the heavy lifting and then provide feedback on how their mixes compare to what they could achieve on their own. That’s what we aim to do—make the process smoother and more accessible for everyone.
Browse the Music AI Archive to find AI Tools for your Music
https://www.musicworks.ai/
Find out more about RoEx
RoEx Links: Website | LinkedIn | Twitter/X | Instagram | Facebook | Discord | Reddit