Automated spoken word analysis will enable the platform to classify songs with greater efficiency, helping users to find more relevant content quicker.
Spotify has successfully patented a “Spoken Word Analyzer” feature, which will allow the platform to make determinations about a song’s lyrical content through the power of AI.
The streaming giant filed the patent, which was ultimately approved by the United States Patent and Trademark Office this week, in late August 2020. The patent describes a feature that would automatically assign a set of tags to songs based on their lyrics. The utility of such a feature will allow Spotify to categorize songs based on their explicitness, mood, and even qualities such as danceability and energy levels, among other criteria.
Tracks have historically had their tags assigned based on user input, but the system Spotify proposes would seek to standardize and automate the practice. While automatic tags are currently prone to inaccuracy, Spotify is incentivized to find operational efficiency in the process due to the sheer volume of content arriving on the platform, an estimated 300,000 new tracks every week.
The feature feels like a logical expansion given the platform’s recent development of lyric-driven search. A superior tagging system will enable users to search for and discover more relevant content based on their moods and stylistic preferences.
Spotify is continuing to innovate increasingly with the help of AI. The company filed a patent to characterize songs based on their “nostalgia metrics” earlier this year, and more recently, the company announced it is delivering an AI-based plagiarism detection system as a new resource for songwriters.
You can read the full “Spoken Words Analyzer” patent filing here.
Source: Digital Music News