Google's Play Music creates playlist based on user's mood

Google's music-subscription service will try to anticipate its listeners' mood swings as it amplifies its competition with Pandora, Spotify and other popular services that play tunes over the Internet.

Starting Tuesday, the $10-a-month All Access service will make music suggestions based on educated guesses about each subscriber's mood and likely activities at certain points in the day or week.

For instance, a subscriber who opens the service on a smartphone on a Monday morning might be offered a playlist suited for commuting, going to the gym or getting motivated for work. Opening the app on Monday evening, though, might generate songs appropriate for eating dinner, studying or unwinding.

Six different music mixes created for different emotions and activities — with such labels as "Jumping Out of Bed" or "In The Lonely Hour" — will be automatically displayed for All Access subscribers in the US and Canada. The mixes won't be played unless the subscriber selects one. The feature won't be immediately available in the other 43 countries where All Access is sold.

The mood music also will be tailored to each listener's tastes, so a subscriber who already has signaled a preference for rock and an aversion for country music would be more likely to hear the Rolling Stones perform "Monkey Man" than "Dead Flowers" in their mixes.

Subscribers also will be able to request playlists designed for specific activities such as napping or housecleaning.

Google's attempt to cater to people's moods reflects the growing importance of delivering soundtracks that suit listeners' discrete tastes and lifestyles. Making the right recommendation is becoming more crucial now that Google, Pandora and Spotify have secured the licensing rights to most of the same music.

"The content is roughly the same, so the main thing you can do for a user now is to have the right context," said Brandon Bilinski, product manager for Google Play Music, which runs All Access. "We want to get our listeners to the right music to fit the mood and make them feel good."

Google Inc. picked up the mood-melding technology in its July purchase of Songza, a free music service with about 5 million listeners.

Google's All Access service launch just 17 months ago, leaving the company that runs the Internet's dominant search engine and other leading digital services in the unfamiliar position of trying to catch up.

Pandora Media Inc.'s free Internet radio station boasts 76 million monthly listeners, while Spotify has 40 million listeners, including more than 10 million subscribers to its $10-per-month service. Google hasn't disclosed how many people subscribe to its All Access service, which offers a music library spanning 30 million titles.

Selecting songs based on listener's shifting moods is similar to what a smart music player called Aether Cone does. That player draws upon the music from another subscription service called Rdio Unlimited, which also charges $10 per month.

Pandora, Spotify and other services all strive to lead their audiences to mixes and genres that will please them, though the others tend to depend on computer algorithms that analyze each person's preferences and listening histories.

Combining human knowledge with a computer's analytical powers is similar to what Beat Electronics was doing with its own music-streaming service before Apple Inc. bought it for $3 billion earlier this year. Apple has said Beats' recommendation system eventually might be blended into its own music-streaming service, though that hasn't happened yet.

Google's new feature includes several thousand playlists assembled by Songza music aficionados that include DJs, performers and critics. Songza's hand-picked playlists will be slightly adjusted by algorithms programmed to learn more about each listener's tastes and habits.

As time goes on, Google hopes to provide even more nuanced playlists that acknowledge a person's mood is likely to be much different while driving to work on a Friday morning than a Monday morning. For now, though, the Mountain View, California, company will depend on cues from each subscriber.

"We can be smart about a lot of things, but it's really hard to tell a person's mood," Bilinski said.