top of page
Image by Daniel Schludi

No, no limits, we’ll reach for the sky

Updated: Jan 20, 2022

Talking to Antoine Sachet – our data science lead – reminded me of the 1993 chart-topping hit No Limits by 2 Unlimited. It features these lyrics….

No, no limits, we'll reach for the sky

No valley too deep, no mountain too high

No, no limits, won't give up the fight

We do what we want, and we do it with pride.

…. and they sum up what Antoine loves about working at SoundOut.

Antoine Sachet - SoundOut
Antoine Sachet - SoundOut

Why did you join SoundOut?

I’m a French engineer with a background in maths. After my Masters in computational statistics and machine learning at University College London, I joined a very large global research company in their development team, but I soon realised that the development process was very siloed. Any new ideas could not be taken from concept to full production without them being passed from one team to the next and the next... It meant innovation was very slow. This really frustrated me, and I decided to join a company that would move faster – give me more of a start-up feeling.

And boy has SoundOut delivered. From day 1 everything has moved so fast. That’s what I love about SoundOut. If I have an idea, I’m given the autonomy to develop it. Here, you’re only limited by how good you are. For example, I added more advanced natural language processing (NLP) to the way we analyse the free-text comments on our world-leading research panel – SliceThePie. I was able to build and test a prototype using the latest open-source models and deploy it in production in only a few months.

Why do you think SoundOut has been able to become the world leader in audio branding testing?

It’s because of the combination of a lot of factors. We have one of the best data collection platforms in the world. This has enabled us to partner with academia to run research projects at tremendous scale which means we understand both the implicit and explicit impact of music. We have a heritage in music testing and continue to work with the world’s leading record labels and artist service platforms. We have a partnership with Professor Daniel Mullensiefen which keeps us on the cutting edge of how to commercialise the psychology of music. And, perhaps most importantly for me, our leadership team embraces rapid innovation.

Where would you like to take the data science next?

Because of the work we have done with Goldsmiths around testing the implicit impact of music and mapping the emotional DNA of music, there are so many exciting ways we can innovate. We are already building a range of tools that draw on our research data. These can be tools that help brands and their advertising and audio branding agencies at every stage in the development lifecycle of sonic brand and music advertising assets.

What excites you about the current Goldsmiths research project?

Last year we carried out a massive research project with Goldsmiths that had over 500,000 responses and produced over six million data points relating to over 200 attributes. This enabled us to create an ‘explicit’ framework for audio branding. Based on that unique data set, we’ve already developed and delivered our BrandMatch tool that helps brands ensure that their sonic logo is perfectly matched to their brand personality.

The latest research with Goldsmiths – supported by the UK government’s Innovate scheme – is even more exciting. This will provide us with an ‘implicit’ framework that maps the subconscious impact of music such as whether some music makes people like a product or a person more or makes a brand feel more welcoming, etc. The implicit power of music is incredible. A recent test we ran showed that some music improved reading comprehension. People who read texts with certain music playing, answered questions about the texts correctly much more often than those who read without the music.

What excites you about the data you have access to?

What excites me most about our approach is that although it is data-driven, it is human-driven too. The six million data points I mentioned before came from tens of thousands of people. It’s the richness of this data that enables us to use cutting- edge models to create new tools. For example, when we developed BrandMatch, we applied a method called Diversity-Induced Self-Representation which enabled us to identify 14 core attributes from the over 200 we tested. So now we have developed a tool that finds the core attributes that are the emotional signature of a music track. This is incredibly exciting for advertisers and brand owners.

Where next for SoundOut’s ‘No Limits’ approach?

Where do I start? There is so much going on. Take voice assistants (VAs) for example. People interact with brands via VAs every day. And this means brands should understand both the explicit and implicit impact of the sound of the voice that is chosen. I’m sure you’ll see something from us on that.

After all, in the words of 2Unlimited

No, no limits, we'll reach for the sky

No valley too deep, no mountain too high!

98 views0 comments


bottom of page