SCA has become an early-stage investor in Australian artificial intelligence (AI) and machine learning company SourseAI.
It will provide content recommendations on the LiSTNR app to continue to improve the user experience.
Consumption of digital audio is growing in Australia and is expected to reach 80% of the population by 2024* and SCA’s LiSTNR is designed to be a world class digital audio listening experience.
SourseAI is a patented augmented intelligence platform that will be used by SCA to discover audience behavioural insights, listener mood states and how to power dynamic cohorts of LiSTNR users, ultimately enabling a hyper-personalised experience across its digital audio platform.
SCA’s in-house analytical capability will be enhanced by SourseAI, which ingests data from SCA’s systems and applies a machine learning algorithm designed to power content recommendations to users. SCA has already connected content metadata including listener topic analysis, from its investment earlier this year in local AI company Sonnant, as well as listening and marketing data to drive data-led decision making within the business.
SCA will also leverage other SourseAI capabilities, such as forecasting and anomaly detection, to understand growth across content genres, predict seasonal content scheduling peaks and troughs, and to explore changing content tastes.
SCA Head of Digital and Innovation, Chris Johnson, says, “The key advantage of developing LiSTNR within SCA is our ability to determine our own future, and we’re leading the market by investing in local, high performance, deep technology partnerships that can accelerate our roadmap. Our investment in SourseAI allows us to deliver a deeply personalised listening experience to consumers, based on consumption habits and context. It also enables us to gain a rich understanding of our audience’s behaviour, enabling us to continue to create market leading advertising solutions.
“The AI and Machine Learning space is scaling rapidly, and we believe that investing in the Australian entrepreneurial ecosystem to support our digital audio ambitions is the best strategy. SourseAI is the right partner to deliver on our vision and our investment will provide significant long-term value to both parties.”
SourseAI CEO, Matt Jones, says, “Sourse’s mission is to apply our patented machine learning algorithms to enhance the experience for LiSTNR audiences using a whole range of data. This includes understanding the tastes and behaviours of every single user on the LiSTNR platform. Then, using our advanced recommendations and personalisation algorithms, we can ensure the right content is surfaced in each individual listener’s feed any time they engage with the app.
“Machine learning models make it possible to drive the key media metrics of frequency of engagement and time spent. By personalising the experience with Sourse, SCA will progress these metrics, resulting in a more engaged audience, and delivering improvement across both advertising targeting and yield.”
*Source: GfK Australian Share of Audio 2019
Subscribe to the radioinfo podcast on these platforms: Acast, Apple iTunes Podcasts, Podtail, Spotify, Google Podcasts, TuneIn, or wherever you get your podcasts.
OK, save them before they go completely and utterly loopy.
It makes sense that IP streaming services use machine learning algorithms in oorder to make models that can tailor a listener's particular music tastes and preferences.
I have made the point before on this site that services such as Spotify may well have access to listeners' preferences, listeners' demographics, artist, genre and country-of-origin. Such streaming services may well be an immediate source of data for analysis unlike terrestrial FTA broadcast receivers.
Such machine learning ('ML') algorithms must be open to sources of all genres. Algorithms must not have any assumption that a person listens to a particular genre. The 'ML' model must be open to all musical genres and open to new entrants in the musical marketplace.
The new entrants may well appear as a suggestive a la the style of a McDonald's suggestive serve of the type "do you want fries with that?" but instead "would you like to preview Bruno Mars and Anderson Paak? Here is a 30 second preview..."
I would also expect that the ML algorithm would also suggest "Puccini: Gianni Schicchi - "O mio babbino caro" sung by Yvonne Kenny accompanied by the MSO.
The listeners' preferences may well provide a more accurate knowledge of listeners' preferences rather than the dictates of radio station management and inaccurate surveys.
I forgot to sign off,
Thank you,
Anthony of exciting and creative Belfield