The mysterious singer with millions of streams - but who (or what) is she? Sienna Rose is having a good month. Three of her dusky, jazz-infused soul songs are in Spotify's Viral Top 50. The most popular, a dreamy ballad called Into The Blue, has been played more than five million times. If she continues on this trajectory, Rose could become one of the year's hottest new stars.

How can you tell if your new favourite artist is a real person? There's a new song doing the rounds, and in the immortal words of Kylie Minogue, you just can't get it out of your head. But what if it was created by a robot, or the artist themself is a product of artificial intelligence (AI)? Do streaming sites have an obligation to label music as AI-generated? And does it even matter, if you like what you hear?

Up to seven out of 10 streams of artificial intelligence-generated music on the Deezer platform are fraudulent, according to the French streaming platform. The company said AI-made music accounts for just 0.5% of streams on the music streaming platform but its analysis shows that fraudsters are behind up to 70% of those streams. AI-generated music is a growing problem on streaming platforms. Fraudsters typically generate revenue on platforms such as Deezer by using bots to "listen" to AI-generated songs – and take the subsequent royalty payments, which become sizeable once spread across multiple tracks. The tactic aims to evade detection measures triggered by vast listening numbers for a small amount of bogus tracks.

The streaming service Deezer heavily relies on the search to help users navigate through its extensive music catalog. Nonetheless, it is primarily designed to find specific items and does not lead directly to a smooth listening experience. We present Text2Playlist, a stand-alone tool that addresses these limitations. Text2Playlist leverages generative AI, music information retrieval and recommendation systems to generate query-specific and personalized playlists, successfully deployed at scale.

This paper examines the influence of recommender systems on local music representation, discussing prior findings from an empirical study on the LFM-2b public dataset. This prior study argued that different recommender systems exhibit algorithmic biases shifting music consumption either towards or against local content. However, LFM-2b users do not reflect the diverse audience of music streaming services. To assess the robustness of this study's conclusions, we conduct a comparative analysis using proprietary listening data from a global music streaming service, which we publicly release alongside this paper. We observe significant differences in local music consumption patterns between our dataset and LFM-2b, suggesting that caution should be exercised when drawing conclusions on local music based solely on LFM-2b. Moreover, we show that the algorithmic biases exhibited in the original work vary in our dataset, and that several unexplored model parameters can significantly influence these biases and affect the study's conclusion on both datasets. Finally, we discuss the complexity of accurately labeling local music, emphasizing the risk of misleading conclusions due to unreliable, biased, or incomplete labels. To encourage further research and ensure reproducibility, we have publicly shared our dataset and code.

This paper presents our recent initiatives to foster the discoverability of new releases on the music streaming service Deezer. After introducing our search and recommendation features dedicated to new releases, we outline our shift from editorial to personalized release suggestions using cold start embeddings and contextual bandits. Backed by online experiments, we discuss the advantages of this shift in terms of recommendation quality and exposure of new releases on the service. Note: This short article presents a work that has been accepted for oral presentation as an "Industry Talk" at the 46th European Conference on Information Retrieval (ECIR 2024).

Two-tower models are a prevalent matching framework for recommendation, which have been widely deployed in industrial applications. The success of two-tower matching attributes to its efficiency in retrieval among a large number of items, since the item tower can be precomputed and used for fast Approximate Nearest Neighbor (ANN) search. However, it suffers two main challenges, including limited feature interaction capability and reduced accuracy in online serving. Existing approaches attempt to design novel late interactions instead of dot products, but they still fail to support complex feature interactions or lose retrieval efficiency. To address these challenges, we propose a new matching paradigm named SparCode, which supports not only sophisticated feature interactions but also efficient retrieval. Specifically, SparCode introduces an all-to-all interaction module to model fine-grained query-item interactions. Besides, we design a discrete code-based sparse inverted index jointly trained with the model to achieve effective and efficient model inference. Extensive experiments have been conducted on open benchmark datasets to demonstrate the superiority of our framework. The results show that SparCode significantly improves the accuracy of candidate item matching while retaining the same level of retrieval efficiency with two-tower models. Our source code will be available at MindSpore/models.

The Venn diagram of HomePod owners and YouTube Music subscribers probably doesn't have a lot of overlap. However, those who use both Apple's speakers and Google's music streaming service may be pleased to learn that the two now play more nicely together. YouTube Music is now available natively on HomePod, meaning that you can ask Siri to play tracks from the service even if your iPhone, iPad or Apple Watch aren't close by. It's now possible to set YouTube Music as the default music service on HomePod. That means you won't have to add "on YouTube Music" when you bark a command at Siri.