In the last few years, artificial intelligence has gone from a novel new piece of tech to an inescapable behemoth of a subject. From boardrooms to bar smoking areas, the applications, implications and ethics of AI are hotly debated issues. While its ability to save time and even lives is undeniable, it’s fair to say that AI’s uses within the creative industries are a tricky topic for some. But as is often the case with new technologies, there are plenty of misconceptions surrounding the use of AI to create audio.
One company with a unique approach to AI within the audio domain is Soundraw. Instead of using AI to approximate what something should sound like based on user prompts, Soundraw uses actual recordings by actual human musicians. We sat down with the Japanese start-up’s Chief Operating Officer, Tao Romera Martinez, to discuss his professional background, his involvement with Soundraw and their refreshing take on AI-powered music production.
‘What's the next biggest fire?’ And then I just go and try to put it out. That's basically my job.
“My name is Tao, I am originally from Spain. I came here as a student and one thing led to another. 20 years later, I'm still here. I've been working in tech, more specifically in early-stage startups for the past 12 years or so. I started my own startup which I ran for about four and a half years. I eventually got a small exit and then kept working on other early-stage startups run by friends and people that I met alongside my own operation. I've been with Soundraw for the past three years as the COO.”
Tao went on to explain more about his role at Soundraw and how he got involved with the company.
“Mostly overseeing anything that needs to be done really. Generally, it's ‘What's the next biggest fire?’ And then I just go and try to put it out. That's basically my job. I'm a product person mostly, so I tend to gravitate toward big projects and tasks around product and growth.
Tao’s clearly experienced when it comes to innovative startups, and we were eager to learn more about his past projects.
“The company I founded was called Tadaku. It was offline cooking lessons which allowed foreigners who were living in Japan to teach their country’s cuisine to Japanese people. You’d all eat together afterwards, so really it was a cover-up for a cultural exchange through cooking.
“I have never been involved in B2B or deep tech until now, it’s always been more consumer-focussed. Although I was working briefly for an English learning app, which was mainly targeted at Japanese and Korean customers. Again, super early-stage startups with less than 10 people in the team. I was also involved with another app service which connected Japanese farmers and producers directly with the consumer.”
Having a sense of purpose in what I do is essential. I’ve always chosen things that I really thought were meaningful for society, for the planet, for us or humankind somehow.
While Tao has a varied background, the thread that runs through all of his past projects is that they are all geared towards making things more accessible and straightforward for users.
“Of course, you always need money to pay the rent, but there are so many things you could do to pay the rent. Having a sense of purpose in what I do is essential. I’ve always chosen things that I really thought were meaningful for society, for the planet, for us or humankind somehow. Not just a job as a way of earning money.”
We asked Tao whether his attraction to Soundraw stems from an interest in music and the tools needed to create it.
“I do have a big love for music. I play the piano and I’m interested in classical music and minimalism. I love techno as well, and I’ve gotten really deep into that, trying to understand the history of music and how music composition works. I’m not a producer or a professional musician, but when I say that I like music, it’s like I really like it.”
We just didn’t see any good coming from not having humans involved in the process of human creativity, for several reasons.
Unlike many other AI-powered music production tools, the samples in Soundraw are originally created by professional producers and musicians. The AI engine then generates a string of variations of those samples, which are combined to create a song within the Soundraw platform.
“Even before services that use generative AI exploded and went viral, such as Suno and Udio, we already had the idea to use real artists and their music. It’s not like we were trying to position ourselves as ‘the good guys’ or ‘the ethical guys’ or anything. We just didn’t see any good coming from not having humans involved in the process of human creativity, for several reasons.
“Of course, there are legal risks that you may incur if you use people’s work to train your AI, but even if it was legal, is it ethical? If you start using people’s work to train your AI engine, people might think that they will suddenly be able to become millionaires by flooding services like Spotify with AI-generated stuff. We didn’t want that to happen either, because what’s the point? Suddenly the whole internet will be flooded with soulless music that no one listens to.
“So for several reasons, we wanted to have humans in the equation one way or another. That’s why since the beginning, our license forbids you to distribute Soundraw music to Spotify or any other DSP unless you modify it after exporting from Soundraw. Of course, we still want to make it easy for people to be able to create music. The whole purpose of Soundraw is to lower the bar so that people who don’t have music skills can still easily make music, but we don’t plan on eliminating humans from that process at all. This is why we also wanted to make it easy for users to modify the stems, change instruments and combine genres.”
Tao goes on to explain who Soundraw is really for.
“We target two types of users right now. There's the creators; video creators or some kind of content creator. And then we have people like artists or music industry professionals who want to use a specific hook in their song, or maybe to create entire beats. These are very different users, but because some of our users have zero music experience, we made it really easy to customize a song with a really simple interface.
“Initially, we did a good job of making Soundraw user-friendly for content creators, but then artists came in and said ‘Actually, we also want to use your music for our songs’. These are a whole different type of users because they do know a lot about music. Not only do they want to be able to customize the stems, but they also want to know the key or chords being used. They’re saying ‘Can I also have more? I like this melody, but I'd like to see some other variations.”
We’re working hard on future versions. You’ll be able to make far more radical changes such as changing the chord progression while keeping the same instruments and the same structure.
Soundraw handles its two main user bases very neatly. For music-illiterate creators, there’s the standard mode. Those who are a little more music savvy can opt to use Pro mode to adjust various parameters of their song including instrumentation, tempo, length, arrangement and even key.
“We’re working hard on future versions. You’ll be able to make far more radical changes such as changing the chord progression while keeping the same instruments and the same structure. All of this came because musicians started showing an interest in Soundraw. Content creators wouldn’t ask for these features, they don’t know what a chord is.
“All of these things allow users to have a pretty granular customization of the song, and that’s not available on other platforms. Other platforms let you input a text prompt, and then out comes a song which you have barely any control over. Maybe you can extend it a bit, you can do a couple of things, but nowhere near the customisation that Soundraw offers.”
It’s not just the creation process that Soundraw wanted to make simpler for their users, it’s everything that comes after that part of the process too.
“We’re very close to launching our distribution service which will allow users to distribute their songs directly through Soundraw with their Soundraw subscription. Artists will be able to not only create beats for their songs, but when it’s finished, they will be able to distribute that song through Soundraw too. This will save users money instead of having to pay a separate service for distribution.”
They trusted Soundraw because of our stance on human creativity.
For anyone reading this who might be wondering where they’ve heard of Soundraw in the past, it’s possibly through their collaborations with artists including French Montana, Trippie Red and Fivio Foreign. Tao shared with us how these kind of collaborations come around.
“It’s a combination of Soundraw looking for partnerships as well as more organic relationships. Those people are already major artists, it’s not like they need the money at all, and they have enough interesting ideas and projects that come their way. So of course, we have access to these people, which is one thing you need to be heard. You need to have some kind of connection to them through our networks and people we know.
“Most importantly, they were generally interested in this kind of collaboration, this kind of project. To them, it’s a cool tool. They trusted Soundraw because of our stance on human creativity. Ultimately, it’s a collaboration so we want to work with the artists rather than either of us working for each other. We want to collaborate with people who like music and technology in general, who like playing with these new tools and seeing what creative ways you can use it.”
We’re talking to artists all the time.
When you’re working directly with some of the music industry’s biggest hitters, you take the opportunity to get feedback and develop your product.
“We’re talking to artists all the time. Just last weekend we had Soundraw Camp happening in New York. There were over 300 artists there doing rap battles and things like that. So we’re constantly getting feedback from these people. All of the features that I mentioned earlier are aimed at musicians, and we’re developing these because artists showed an interest in Soundraw. They began giving feedback about things the platform was missing.”
In terms of artist-focussed features, we have so many things coming up.
With that in mind, we asked Tao if there are any Soundraw features currently in the pipeline that he was able to share with us.
“For API customers, definitely the video-to-music, which is going to be amazing. It will allow users to come to let’s say Filmora or somewhere, to quickly edit their video. You pick a genre and length, and it will analyze the video and then generate a few suggested songs. The current video-to-music tools have a few different problems. First, if you need to provide a prompt, you need to explain very clearly what is happening in the video.
“This might be even harder if English is not your native language. Also, if you’re not into music, you might not know what type of mood or vibe of music you want for your video, you might not know what Funk sounds like! To avoid all of these issues, we’re very close to allowing users to be able to just upload their video, we analyze it and then provide a few options.
“In terms of artist-focussed features, we have so many things coming up. There’s the music distribution which is coming really soon, plus a bunch of updates to the user interface and experience. Things like how you can control and adjust the stems and instruments. Then of course being able to see and change the chords being used in the song, adding stem variations and having access to more instruments.
“We’re also completing a major overhaul of the way songs are created. One of the most common pieces of feedback we receive is that songs start to sound repetitive when they’re long. This can happen depending on the combination of tags that are chosen to generate the song, so we’re working very aggressively to solve that.”
It’s clear that Tao and Soundraw have a distinctive philosophy when it comes to developing useful AI-powered music creation tools. They’re aware of AI’s limitations in creative fields, but they aren’t trying to reinvent the wheel of music production. They’re simply exploring new ways of using the available technology to spark inspiration in creators and make their lives easier.
