News

These are the latest developments in our project and the ABC_DJ consortium in general.

HearDis! & ABC_Dj at Sound Stockholm 2018


HearDis! has been invited to talk at SoundStockholm – the conference that this November will tackle the growing audio consumption, particularly how people are influenced by music and how brands can use sound together with new technologies to achieve their marketing goals! In preparation for the event, Resumé – largest Swedish magazine about media and marketing – interviewed HearDis! to get a sneak peek on its audio branding approach and the latest technologies developed through our ABC_DJ project:

Could you give some examples of how factors as brand message, core values, target group and brand architecture affects the music/sound choice?
Any given brand identity can be mapped into a list of 36 semantic terms (e.g., “cool”, “urban”, “happy”, etc.) which our algorithm then uses to predict the musical perception of such terms among 40 different target groups (e.g., track XX is 89% cool, 76% urban 64% happy). Consumers are allocated in 9 Sinus-Meta-Milieus (e.g., “established”, “sensation-oriented”) and 31 demographic groups according to education and age. The above-mentioned list of 36 semantic terms – which we refer to as General Music Branding Inventory (GMBI) – was created empirically in an interdisciplinary workshop with international experts from marketing, audio branding, record labels and music publishers. It was furthermore refined and evaluated in a large-scale survey with international marketing professionals.


Does your mood-filter work similarly to, as an example, Spotify’s mood-based auto-generated playlists?
Today several companies adopt moods filters for music search, discovery, recommendation or to automatically generate playlists. Spotify for instance uses big data to feed its algorithms. Instead, we are the only company on the market to offer an algorithm that interconnects ground truth data from listening experiments directly with consumers (i.e., 10,000+ participants from different countries), associated demographic data, the Sinus Milieus from the social and market research company Integral Marktforschung, and a large amount of audio features. Thus our algorithm is capable of predicting the musical perception of any given song within any given target group solely based on the audio features of the sound recording – an unprecedented and unique approach.
Since there is no data available of evaluation activities from e.g. Spotify or similar companies, it is not possible to get an objective perspective on the power of such algorithms. Our algorithm, in contrast, has been tested on accuracy and predicts the perceived musical meaning with a success rate of 80.1 %.

Has there been any kinds of complications with the algorithms, and how have you fine-tuned the algorithms and the music selection process?
Sure, we experienced many obstacles and complications when developing the algorithm. First, we uncovered a fitting factor model capable of being transferred to the regression model; we defined the audio features extracted from sound recordings that fed the algorithm; we created accurate training samples for music information retrieval and finally match the properties list to real world issues, such as the speed of the algorithm and related accuracy. It was indeed a laborious process with many unforeseen complications yet we are now proud to have reached a prediction accuracy currently set at 80,1%. Additionally, the algorithm is continuously learning over time with the feedback from our music consultants.

Are you the only agency offering this type algorithm-based music branding service?
Indeed we are! We are the first one to have developed an algorithm to specifically support and boost audio branding efforts. This means we are pioneers for having merged know-hows across Music and Marketing (from both a brand and a customer’s perspective), making the resulting information actionable through the latest Technology: our AI incorporates data capable of predicting brand-fit music based on acoustical features of music and socio-cultural context of listeners – which positions HearDis! at the centre of an Audio Branding revolution! Since first raising awareness about our R&D on this front at worldwide industry gatherings and to the public, other agencies have gotten inspired and are attempting to adapt our technology and integrate it into their own services. However, we know that so far those approaches are merely using a fraction of the overall ground truth we base our AI on.

Also, it is important to note that we are alone in serving all touchpoints a brand may be concerned of, be it TV, online or retail. Companies such as Soundtrack Your Brand, for instance, take care of the retail aspect only while other audio branding agencies focus on handling TV and online requests. We rather advocate for a 360 degree approach, which establishes us as the full service agency that our clients depend on for any music- or audio-related matter in their marketing strategy.

HearDis! Website

Is there a growing demand for audio branding services, and do think it will develop and be more and more common in the future?
Definitely. Audio Branding is a growing business segment with huge potential for the future, applicable to an extensive range of settings, from a more traditional fashion showroom pursuing the optimal atmosphere to showcase the latest additions, to a more futuristic electric vehicle needing sophisticated sound design to embody and manifest its brand.

In order to succeed in an ever-increasingly competitive market, brands need to stand out from the pack – for which a strong corporate identity is necessary. Because of its emotional power, Music plays a key role in marketing strategies, in that it can subtly represent brand values and effectively convey these to the respective target audience.

This becomes more and more relevant as we witness the evolution of commerce. An example can be found in the retail industry, which is experiencing a disruptive transformation, where the point of sale (POS) turns into a point of experience (POX): whichever the retail space – pop-up store, showroom, trade fair booth, etc. – music is a valuable asset to complement and enhance the multisensory experience to elicit customer engagement and increment brand perception.

Another example is the automotive industry. As we advance towards a greener future, car manufactures need to come up with alternative ways to personalise their products: until now, the specific engine sound serves as the brand’s signature. Yet what is to replace this when vehicles of tomorrow are electric? How can a car manufacturer make their electric vehicles sound distinct and provide a unique (brand) experience? By developing cutting-edge forms of audio branding, the automotive sector surely offers great opportunities that we are already tackling.

 

 

 

 

Company Description: HearDis! is a full service agency for corporate sound founded in 2005, with offices in Berlin and Stuttgart, Germany. HearDis! mission is to serve companies with authentic and tailor-made Audio Branding solutions to craft the optimal corporate sound that makes brands audibly distinctive and memorable. Since 2016, HearDis! is part of ABC_DJ, which engineered an algorithm to predict brand-fit music.

At the intersection of Music, Marketing and Artificial Intelligence, HearDis! is today at the forefront of disruptive technologies that revolutionize the realm of Audio Branding as we know it. Clients include renowned companies such as Hugo Boss, Mercedes-Benz, Mercedes-AMG, Vitra, Mykita, Birkenstock, Porsche Design, GANT, KaDeWe Group, Monkey 47, Motel One and many more.

Recap: The New Shape of Audio Branding

This slideshow requires JavaScript.

Paris is always a special city yet it was extra special on the recent ABC_DJ workshop during ISMIR conference. At IRCAM‘s premises, we performed the very first live demo of the advanced tools we engineered over the last two and a half years. We tested our brand-fit music algorithm with audio design agency Sixième Son on its client – French multinational automaker Renault – which revealed impressive results:

This case study really showed the power of ABC_DJ. We have checked the original brand song of Renault. The algorithm says it’s a perfect (brand-) fit. This solution is truly valuable, for instance when agencies need a proof of concept for a brand.” (Andreas Schoenrock, Musicologist, HearDis!).

Sixième Son’s Managing Director Laurent Cochini was also inspired by ABC_DJ’s potential and affirmed that

From now on, the challenge won’t be to deliver Audio Branding. The challenge will be to deliver THE RIGHT ONE“.

Such groundbreaking solutions have been made possible thanks to the European Commission’s Horizon 2020 Research and Innovation programme and the support of our consortium partners such as HearDis!, Integral-Marktforschung, TU Berlin, IRCAM, Fincons, Fratelli Piacenza and Lovemonk.

Stay tuned to find out how ABC_DJ’s disruptive technologies are to impact the realm of Audio Branding as we know it.

C/O Pop Convention

 

 

 

 

 

 

 

As part of the upcoming C/O Pop Convention, HearDis! will partake to the special Music Unchained edition on the Future of Synch, untapped opportunities, and challenges ahead for the AV industry – curated this Friday in collaboration with MusicTech Germany

CEO Robin Hofmann has been invited in the expert roundtable, where Music Unchained aims to facilitate the dialogue between the different protagonists of an ever-growing creative community, elaborating on ideas, best practices, new technologies and media formats, and the overall potential for collaboration.

Audio Branding Workshop

On September 20th, HearDis! and Sixième Son will host “The New Shape of Audio Branding”, a half day workshop at IRCAM in Paris. 

The workshop will tackle the current challenges to bring Audio Branding into the digital future, addressing topics in scientific research of music information retrieval and music psychology presenting the advances made in our ABC_DJ project.

“The New Shape of Audio Branding” is one of this years’ five satellite workshops which will take place around ISMIR conference in Paris.

Register here (limited capacities!):

www.audiobrandingworkshop.com

ABC_DJ @Sound and Music Computing, Cyprus

 

 

 

 

 

We’re glad to announce that our publication “Heuristic Algorithm for DJ Cue Point Estimation” will be part at the Sound and Music Computing Conference in Cyprus. The main theme of the conference focuses on the ability of sound and music to cross boundaries, to eliminate borderlines and overcome physical and digital limitations.

Our Dj Cue Point Module is an automatic track annotation and DJ mixing algorithm which was developed in the context of audio branding for in-store music delivery.

This publication was jointly realised by IRCAM and HearDis!.

 

ABC_DJ at Music Information Retrieval Berlin Meetup

 

 

 

 

 

On June 25th 2018 we present our knowledge-based music branding recommender system at MIR Berlin Meetup.  The ABC_DJ recommender system requirements significantly differ from traditional music recommenders: In our case, the perceived semantic expression of music titles is of main interest since it has to meet marketing intentions, whereas consumers’ personal preferences or emotional responses are of rather minor importance.

In order to address the ‘semantic gap’ between audio signal analysis and complex brand identities to be communicated by music to heterogeneous target groups, our system combines machine learning of music branding expert knowledge with audio signal analysis toolboxes’ output and population-representative ground truth data gathered by multinational online listening experiments.

The MIR Berlin Meetup will take place at The Hybrid Lab a work, event and experimentation space serving as cross-disciplinary exchange of art, science and technology.

“Signal Processing and AI: Changing the World”

Glad to announce the release of the paper “Fast and adaptive blind audio source separation using recursive Levenberg-Marquardt synchrosqueezing” by IRCAM‘s Dominique Fourer and Geoffroy Peeters during the highly acclaimed International Conference on Acoustics, Speech and Signal Processing (IEEE ICASSP) in Calgary, Canada.

This paper revisits the Degenerate Unmixing Estimation Technique (DUET) for blind audio separation of an arbitrary number of sources given two mixtures through a recursively computed and adaptive time-frequency representation.

Any more questions? Dig deeper and visit the IEEE ICASSP next week. Or wait for the paper download, soon to be released on our Publications site

Save the date: “The New Shape of Audio Branding” workshop

ismir2018logo_black_long
The ISMIR conference is held annually and is the world’s leading conference for Music Information Retrieval (MIR). It will take place in Paris, France, from September 23-27, 2018. This years ISMIR is a join organization between Télécom ParisTech and IRCAM. We’re happy to announce to held the workshop “The New Shape of Audio Branding” as part of ISMIR 2018 satellite events at IRCAM on Thursday 20/09/2018.

Subscribe to the ABC DJ Newsletter for more info.