Spotify is the world’s most popular on-demand audio streaming company, with over 299 million active users listening to content from our catalog of millions of tracks and podcasts. Discovery and personalization play a key role in serving the right content to our users at every moment. With increased adoption of voice assistants and devices, voice presents a new paradigm to provide our users a friction-free and a personalized experience. Additionally the lack of visual interface in Voice clients introduces new challenges and exciting opportunities. In this talk, I will discuss how we apply personalization to our Voice stack to enhance our users’ experience. The ability to help users easily access the right content with limited input becomes extremely high utility for the Voice use cases where the users’ utterances are freestyle natural language and range in intent from being super specific to ambiguous. Finally, we will also discuss some of the unique challenges in applying personalization techniques and evaluating impact in Voice and our future work.
Session Summary
Play Spotify
MLconf Online 2020
Vidhya Murali
Spotify
Machine Learning Engineering Manager
Learn more »