Hey Spotify, What Can I Ask You?
A quick guide on what features are available with the new “Hey Spotify” voice experience
On April 9, 2021, Spotify announced the release of a new hands-free voice-controlled experience. It’s called “Hey Spotify” and it’s designed to listen to your every request. For the most part.
If you’re familiar with the format of the voice-controlled experiences of Alexa, Google, or Siri, this is pretty similar — the main difference being that Spotify listens to you only when you have the app open on your phone. Same as with the other big voice assistants, Spotify’s voice controls policy outlines that the data being used in your voice interactions will only be used to make the assistant smarter, that is, to make sure the wake-word actually wakes up the feature and to improve upon music recommendations.
What that means for you is if you opt-in to the feature, Spotify will record “short snippets of a few seconds which are deleted if the wake-word is not detected.” This is not to be a matter of alarm and is, in fact, the way all voice assistants tend to be trained. Much of what makes a smart voice assistant smart is the ability to learn from live customer data. After all, how would machines better understand us if they’re not taught with real-time “first-hand” experience? It’s like asking an English-native speaker to learn textbook Japanese without ever hearing a fluent Japanese speaker say anything out loud.
To turn on the feature and customize your assistant experience, the easiest way is to go into your settings and scroll down past Storage and Notifications until you reach the “Voice” section. After you give microphone permissions to the Spotify mobile app, you’ll be able to tap the toggle to enable “Hey Spotify”. There are currently two assistant voices available. Mine defaulted to Option 2, which is a more feminine voice, while Option 1 goes for a more traditionally masculine voice.
I will say that, as a Conversation Designer myself, the “Hey Spotify” feature is pretty brilliant and already adept at handling various types of voice-initiated queries beyond the “play music” command.
What “Hey Spotify” Can Do
Regardless of where you are on the app, “Hey Spotify” will pull up the Spotify assistant and the associated modal screen. It should be noted though, most of the current prompts and functions highlight the music playing capabilities of the experience. If you ask “Hey Spotify” straight-up “What can you do?”, the assistant replies with: “Ask me to play any artist, album, song, playlist, or podcast. For example, you can say, ‘Play today’s Top Hits’.” This focus is not bad per se. It’s meant to show how easily Spotify can be integrated with hands-free activities, like wanting to stream music while driving. At the present, “Hey Spotify” is not a conversational experience.
As far as other features, here’s everything I tested:
- Find and play songs by name and artist
- Find and play a playlist by name (it didn’t recognize me saying the word “playlist” the first time around)
- Skip ahead to the next song in a playlist
- Skip back one song in a playlist
- Remove and add “shuffle”
- Remove and add “on repeat”
- Pause and unpause a song currently playing
After processing a voice command, visual and audio (earcon) cues will confirm that a query has been understood.
Where “Hey Spotify” Currently Falls Short
Surprisingly, asking Spotify to “play something I like” did not automatically trigger the app to start playing my Liked Songs— it instead defaulted to a random Daily Mix (which I never listen to). For its first iteration, I found the overall Spotify voice experience enjoyable, albeit mildly frustrating.
Spotify can’t hear you if you’re playing music out loud on your phone speaker and trying to talk to it at the same time. Unlike the Google Assistant, it doesn’t attempt to momentarily pause the music either. Additionally, when I tried to record my phone screen for a demo of the assistant’s voice and reprompts, it refused to wake up entirely, as the screen capture settings of my phone were also using my microphone.
Regarding the technical limitations of the experience: I believe there are plenty more queries that should and will probably be added. Part of what makes Spotify so memorable and a formidable product in the market is its tailored UX. It’s an app that’s easy to use and highly personalized. Just like how it’s context-aware in the greetings it shows on the home screen (“Good evening” during the later parts of the day), so should “Hey Spotify” strive for context-awareness.
Just like how [Spotify is] context-aware in the greetings it shows on the home screen, so should “Hey Spotify” strive for context-awareness.
As a user, I want to be able to ask Spotify to play the last thing I heard or at least take me to a page where I can see my “Recently Played”. I don’t want it to try to extract a song title from what I say. When in doubt, the assistant attempts to find songs that match some or all of the words you say whenever you ask things in a format it doesn’t understand. If it can’t understand you at all, the “Try another request” screen shows up time and again, with no multi-layering of error messages or fallback reprompts. It’s simply a “Sorry, I didn’t understand you” error message by another name and gets annoying pretty fast.
Still, I’m excited to see the product grow. As a #VoiceFirst enthusiast, I’ll definitely be keeping an ear out for any new developments from the Spotify Voice teams.
Edit: on April 13, 2021, Spotify released a limited release physical product, “Car Thing,” which is meant to be the in-car interface for the in-app only experience “Hey Spotify”.