Musiio provides AI-powered analysis, tagging and search tools to some of the world’s largest music catalogs, counting Sony Music, Hipgnosis, Amanotes, Epidemic Sound and Blanco Y Negro among their customers.
Avid rock guitarist turned co-founder and CEO, Hazel Savage spent 15 years in the music industry working for some of the world’s biggest music brands – from stacking shelves at HMV to running teams in companies at the forefront of music listening and recommendations, Hazel understands the needs of the industry, from the musician to the large multinational.
You have been in the music industry for over 15 years, what makes you so passionate about music and why did you want to get involved in the music industry?
My parents were pretty rock and roll. They were huge music fans, so I was always surrounded by music growing up. Then, for my 13th birthday, I got a guitar. I still play and I have a passion for live. So when I figured out what I was going to do with my life, it made sense to focus on something that I had been devoting most of my time to.
I ended up doing lots of tangentially related things. I played in a band. I managed groups. I organized club nights. I was handing out flyers for other people’s club nights, keeping guest lists, and before I knew it, it turned into a career, albeit certainly with a technology focus.
Could you share the genesis story behind Musiio?
My first job after university was stacking the shelves at HMV (the UK record store). So, you could say that I’ve been aware of the music categorization issues ever since. Fast forward a few years (via Shazam, Pandora, and Universal), and I was working for a UGC music platform with thousands of tracks uploaded daily. I worked with a playlister who had to manually collect top music downloads into playlists. He listened to hundreds of songs a day. Some days he had enough suitable content for a playlist. Some days he didn’t. I started wondering if there could be a way to automate finding the best leads for a given scenario. That way he could use his skills as a music expert for curation, rather than just acting as a filter for bad music.
Musiio was formed when I met my co-founder Aron Pettersson through the Entrepreneur First start-up incubator in Singapore in 2018. Aron is an AI whiz. When we were talking about ways we could work together, we realized we might be able to use Aron’s AI skills to solve the problem of music-based filtering, tagging, or automatic music search. with genres, moods, BPM, etc. or fingerprint-based searches. . Aron built a prototype of the algorithm in an afternoon, and we ran it on a free music archive. We went out to lunch, letting him process the data. On our return, we were amazed by the precision of the results. We couldn’t have hoped for a more successful proof of concept. From there, we massively optimized the algorithm. We have a music team that helps teach AI and does quality assurance, and we’ve launched products for tagging, finding audio references, playlisting, and even selecting song segments for platforms. such as TikTok.
What are the different types of machine learning algorithms used?
We have built our own proprietary algorithms, and we consider this our secret sauce! My co-founder Aron has been at the forefront of machine learning for over a decade, working in molecular biology, neuroscience, physics, and even game development. He leads our AI team. We also leverage the great technologies available such as TensorFlow, Kubernetes and Google Cloud Services for scalability and to deliver our products at scale, at our highest volume we were scoring 5,000,000 leads per day! We’ve also spent a lot of time and effort streamlining our workflows in JIRA; it’s not just about the tools you use, but how effectively you can work with a team of developers and music experts. The association of the two AI and Music teams is the second part of our secret sauce.
What are some of the challenges of creating a search engine for music?
Speed and accuracy are the great challenges of research. It must be fast because people use it in real time. This is different from markup because a user will often perform multiple search queries, but markup only occurs once.
There are several things you can do to speed up the search. You can just show tracks that share the same tags as your reference track, but you’ll sacrifice accuracy. A pure audio benchmark search through a catalog of 200 million tracks, for example, can be time-consuming, so you constantly have to balance speed and accuracy and search for solutions. It’s tricky and some of it is hard-earned knowledge, but what I can share is that we convert audio files into spectrograms, very detailed fingerprints of audio files and when we do reference research audio, the algorithm analyzes up to 1,500 data points – far beyond what is possible with keywords alone. And it picks up hard-to-describe musical characteristics such as vocal quality, mood, and mood. We also allow users to set filters, so their searches can be faster and more targeted.
Another challenge is how you manage relevance. Most people don’t get past the first page of results, so we spent a lot of time on that.
What problems does Musiio solve for b2b customers?
We serve anyone with a catalog of music. We’ve built the technology to scale whether you’re a musician who doesn’t have time to tag music and wants to focus on creating, or a streaming service with hundreds of millions of tracks.
We help record labels organize their data for better catalog browsing, we help sync companies (who put music on video/TV and film) discover hidden gems, and we help services streaming to create better playlists. The problem that all of these companies face is that manually processing the audio while listening to each track is labor intensive and difficult to do accurately for an extended period of time. I tagged 1000 tracks as an experiment. It took two weeks and it wasn’t fun at all. Our AI can tag millions of leads per day with 90-99% accuracy.
With our Musiio Search product, we enable our B2B customers to offer audio reference search as a feature. If a video producer is looking for music placement, they will start by understanding their client’s expectations for genre, mood, BPM, and then research the site of their choice.
Musiio shortens this process with our partners who install our search by allowing the same video producer to use a “reference track” to search the entire database in seconds. Our AI will scan the reference track and return the closest audio matches.
Musiio recently launched an NFT Song Slicer product, could you describe what it is?
NFT Song Slicer is a prototype designed to help artists get the most out of their music. It uses an AI-driven process to find desirable hooks in a track — up to three per song — and give timecodes so an artist can create those song sections as NFTs. It can also do this automatically for entire catalogs, allowing labels and artists with large catalogs to quickly create new collectible digital assets.
What are the potential use cases for this type of Song Slicer product?
For catalog owners or artists with a large catalog, NFT Song Slicer can select the most valuable sections from millions of songs per day. Record labels, for example, can then turn those slices of songs into NFTs and sell them as limited-edition digital products.
With the streaming revolution, it has become difficult for fans to have a dollar in their pocket from the artists they love. We view NFT Song Slicer as a way for fans to support their favorite artists and for fans to own digital collectibles. Each band can also be priced differently by a rights holder. For example, a chorus may cost more than a verse.
And, because NFT Song Slicer identifies the most valuable sections of a track, we see this technology deliver value predictions for NFTs and even entire music catalogs.
What is your vision of the future of Musiio?
I say Musiio is one third of a billion dollar company. To build this business, you need three parties. The first is lawful access to large volumes of data, or a “pipeline”. The second part is technology. It’s us, and we’re very good at what we do. The third and final part is a label: a way to monetize what you find, search for, or discover. Musiio is still working towards this long-term goal.
Do you think AI will be able to write and generate music in the near future?
I’m pretty candid about not being a big fan of AI for creativity. It’s a fun academic experience, and there are systems that do it, but I just don’t see the need for it. Musiio works so well because nobody wants to tag thousands of songs a day. It’s not fun and you don’t need a person to do it efficiently or quickly. But making music? I’m not so sure. There is no shortage of people who want to make music.
Even so, I think we’re at least five to 10 years away from AI-generated music that sounds great. I heard some AI-generated piano music the other day, and it’s hard to tell if it’s AI-written or just someone who isn’t very accomplished. I’m not convinced that an AI performance will ever be indistinguishable from an accomplished human gamer.
And why would you want it to be? A big part of what makes music interesting is the lore around an artist, their personality, style, and message. It’s not just about the music.
Is there anything else you would like to share about Musiio?
I’m very happy that Musiio was awarded fourth place in Fast Company’s 10 Most Innovative Music Companies of 2022. Our team and technology have grown from the seed of an idea to international recognition alongside huge names in industry such as Hipgnosis and SoundCloud. It’s a tribute to the blood, sweat and tears our team put into our industry-leading products. We are excited to be at the forefront of the intersection between music and technology. And knowing that there are use cases we haven’t even thought of yet makes me very excited for the future.
Thank you for this excellent interview, readers who want to know more should visit Musiio.