Several platforms have recently developed, through their algorithms, musical services intended to highlight emerging talents. A use that questions the functioning of these new technologies, as well as the evolution of platforms and their role in the music industry.
The development of the Internet has changed our access to culture and our consumption. New digital methods have appeared, whether in the field of cinema, with video streaming platforms (Netflix, OCS, Disney+, etc.), or in that of music, with Spotify, Deezer or Soundcloud. and Pandora. Playlists created automatically using recommendation algorithms have replaced CDs, while these applications have, more generally, upset creation, production and listening practices.
But these music streaming platforms are also trying to find their place in the market for new talent. 1er March 2022, Spotify filed a patent for an emerging artist detection algorithm. For its part, the social network TikTok has created SoundOn, a service to promote performers who are still little known, while SoundCloud has announced the launch of its stable project dedicated to the musicians of tomorrow.
Origins and operation of recommendation algorithms
Recommendation algorithms are not new. With the development of new technologies, streaming services find themselves at the heart of a paradigm in which all audio data is located on servers. This storage then allows their analysis, whether by the way people interact with the content, or by the content itself.
The first audio service to have implemented such an algorithmic analysis system is Pandora. This is explained by Geoffroy Peeters, professor at Télécom Paris: “Pandora had annotators who analyzed each piece of music according to 400 extremely fine and relevant criteria to describe the handmade music. » After which many actors who work in the removal of musical information have developed algorithms which make it possible to analyze the content of a piece and to extract its characteristics. This system allows music platforms to establish a catalog and recommend content to each listener based on acoustic criteria.
Alongside these content-specific characteristics, social and environmental characteristics must also be taken into account in the final analysis. “Technologies certainly make it possible to recommend songs that have the same characteristics to a user, but other interfering parameters must also be taken into account. People don’t want to listen to the same type of music all the time, because we would go in circles. »
From content analysis technologies, there are therefore different ways of creating suggestions, depending on how the content part and the environmental and social part are weighted. This global scheme gives rise to new recommendations, and thus allows the discovery of new talents.
There is also a dose of surprise, in particular thanks to playlists, a model widely adopted by Pandora, Spotify and Deezer. “Platforms have quickly realized that people mainly listen to playlists. Where Spotify and Deezer offered unlimited listening to 60 million tracks, they quickly returned to a radio system – like Pandora – by offering more playlists created automatically, based on recommendations”explains Geoffroy Peeters.
The evolution of music platforms
This system of playlists is one of the first markers of the evolution of music platforms. But the latter went further, concerning emerging artists, often placed at the end of the list and therefore rarely highlighted. Recommender systems thus suggest titles in order to solve the problem of statistical distribution. Geoffroy Peeters details: “Although 60 million tracks are potentially listenable on Spotify and Deezer, users will ultimately focus on more or less 1,000 tracks. Thus, to allow the highlighting of titles that are never listened to, we will offer users a title similar to those listened to in terms of vocal characteristics, placed at the end of the playlist. We will be able, from the content, to put it forward and re-offer it to the user. This is really the main motivation of the platforms. »
This functioning has several consequences for the actors of the music. In addition to artistic diversity for the user, record companies can thus push their “back catalogue”. On their side, the emerging artists, although there is a strong competition to integrate these playlists – what is more those of the front page – will benefit from wider distribution on the platform.
“Spotify has become the preferred transmission channel for artists. »
professor at Telecom Paris
This is what Charles Picasso, artist and engineer at Ircam (Institute for Research and Acoustic/Music Coordination) defends: “In my opinion, it’s attractive, just from the point of view of content delivery and quick visibility. Everyone has access to Deezer or Spotify. » This wide distribution will also go through the provision of new services by music platforms – a definite contribution for these new talents. For example, Pandora set up a service about ten years ago to connect artists with their listeners.
“The economic model based on monthly subscription essentially finances the remuneration of big artists and we don’t have a viable model for emerging talent. »
artist and engineer at Ircam
Spotify, meanwhile, has become a true content creator. “Spotify has become the preferred transmission channel for artistsnotes Geoffroy Peeters. The platform has gone further, becoming a content creator with the Spotify Artists, Spotify Podcasts and Spotify Soundtrack services. It is a virtual recording studio in which the artist will create the music himself using Spotify tools. Then, he will directly sign and distribute his music through the platform, and be put in touch with his audience. » The platform uses very advanced artificial intelligence tools to help the artist compose. A recent innovation, which did not exist when we started to predict musical content automatically. The application of new algorithms is therefore not only intended to detect new talents, but also to provide them with creative and production tools.
New technologies and platform algorithms: competition for record companies?
Although matchmaking and creation services have evolved on the platforms, the arrival of algorithms to identify new talent does not represent direct competition with record companies, as long as they do not offer a very identified service. , according to Charles Picasso. There is also the question of the remuneration of these new talents, once unearthed. “Currently, the economic model does not help independent artists at all. It only boosts mainstream listening in terms of visibility, because visibility matters and is going to be placed on artists that perform better. The economic model based on the monthly subscription essentially finances the remuneration of the big artists and there is no model that is neither intermediary nor viable for emerging talents. »
“The majors are not at all in technologies and artificial intelligence. »
professor at Telecom Paris
In addition, record companies have significant leverage in terms of visibility, negotiations and copyrights. They have always discussed with the platforms, and as long as a real movement of artists that self-produces does not emerge, or that the platforms do not offer a real market alternative, the creation of these algorithms does not seem to encroach on the room for maneuver of the majors and labels. A point of view shared by Geoffroy Peeters, for whom the two systems are brought to coexist: “The majors are not at all in technologies and artificial intelligence. So I think it can be complementary and I don’t know if these search algorithms will really replace them. I can hardly see anyone with an older musical tradition benefiting from this kind of technique. So the two will necessarily coexist. »
However, the professor does not ignore the question of intentional short-circuiting by platforms like Spotify, thanks to these new detection and creation tools. It is true that signing with a record company can be a brake on creativity, given that the majors are only open to recognized talents. In this case, platform services and the arrival of new algorithms to highlight lesser-known artists appear to be a dream solution for them.
The new technologies have undoubtedly upset the music industry, both in terms of distribution and creation. The era of bargain hunting at the record store is definitely over. Now, algorithms optimize listener searches as well as the visibility of new artists. The use of these new technologies by audio platforms has also encouraged the development of their services. A system that could be perceived at first glance as unfair competition for record companies and labels vis-à-vis emerging talent, but which tends more to coexist with them than to replace them.
Be that as it may, this model questions more broadly the influence that technologies can exert on musical creation. Is there a risk of entering into a homogenization of a model by following the algorithmic codes? How will artists position themselves in the digital art market, at a time when NFTs and the metaverse are emerging? After the recommendation and detection algorithms, these new data are in any case the reflection of a permanent race – both for artists and for all other players in the music industry.
We would like to say thanks to the writer of this post for this outstanding content
Music platforms, unearthing new talents?
We have our social media profiles here and other related pages herehttps://www.ai-magazine.com/related-pages/