Revolutionizing Music Discovery: Aria in Amazon Music
Tailoring Every Beat to Your Mood and Preferences.
Introducing Aria - AI’s Symphony
Aria leverages AI-driven personalization and deep analysis of user habits to curate tailored recommendations, adapting dynamically to user interactions for unique music discovery experiences.
Empathizing
We choose 3 different type of research methodology to fully understand our user group. Gain insights on both generative AI and music streaming services, quality and quantity.
01
User Interviews
Get a comprehensive understanding of our users' current music experience through moderated interviews. Establish a basic concept and collect major pain points and needs.
02
Expert interviews
Conducted in-depth discussions with music industry experts to gain insights into user behavior and preferences in music streaming.
03
User survey
Sending surveys to understand what preferences our users have when it comes to using AI. This is to ensure that when we add AI functions, users' task flow can appear in a form they are familiar with.
5
Designed with Users in Mind
- Aria’s User Centric Approach
We use affinity mapping to collect user’s thoughts and major needs they have.
6
Key User Insights
We conducted 12 user interviews to learn current user pain points and their potential needs.
“Desires a more intuitive way to create and manage playlists. With the specific songs from the playlist which was created.”
“Need a more intuitive and engaging way to discover music similar to current favorites and an easier method to create and manage playlists.”
“I want a easy way to create personal playlists and the importance of an intuitive, user-friendly interface.”
7
The Problem
Challenge #1
Inefficient Borrowing Process
Expert Interviews:
Bridging the User Insight Gap
“Leveraging tech to tap into listeners' preferences isn't just smart—it's very important. It helps us deliver music that speaks directly to each person's soul.”
”Listening intently to our audience's favorites and exploring their distinctive tastes is key. That's how we create playlists that resonate as if they were personally curated for each listener.“
60%
Comfort with AI Recommendations
44%
Familiar/ Aware of AI features
47%
Daily Interaction with AI
Deepening Understanding through Surveys: Extending Insights from User Interviews.
Conducted surveys with 36 users to deepen our understanding of AI experiences, following initial
user insights.
Are Users Ready for AI in Music Streaming?
9
9
Key Takeaways
Discover
Add more discovery possibilities for users.
Easy interaction with few steps to operate.
The results of discovery can be highly personalized.
Personalization
Easy and quick
11
Our Problem Statement
How might we design a feature for Amazon Music to provide users with a seamless and personalized music discovery, aligning with their specific tastes and daily activities?
How might we design a feature for Amazon Music to provide users with a seamless and personalized music discovery, aligning with their specific tastes and daily activities?
13
"As a young music listener trying to discover new music that fits my mood, genre, and activities, I really don’t like getting repetitive playlists because it lacks variety and personalization."
"As a music enthusiast, I need an intuitive and dynamic
recommendation system that adapts to my
specific tastes so that I can enjoy more personalized
music during my daily routines."
What Are User Pain Points and Needs?
14
"I want music that adapts to my mood and elevates my day without endless searching —no app has satisfied this yet."
Goals
Mood Alignment: Instantly access music that mirrors current emotions, enhancing immersion.
Discover New Sounds: Dive into new genres and artists that resonate with personal tastes.
Integrate Music: Effortlessly blend music with work and exercise routines.
Navigating vast music libraries to find mood-aligned tracks.
Encountering repetitive playlists that lack personalization.
The inconvenience of manually curating playlists to match diverse preferences and activities.
Frustrations
Motivations
Emotional Resonance: Seeks music that reflects and enhances current emotions.
Self-Expression: Uses music to articulate personal identity and mood.
Personalized Discovery: Prefers an intuitive system for finding new music that matches mood and context.
Jonah, a data analyst, uses music to fuel his focus and creativity. Frustrated by repetitive playlists, he seeks a service that customizes music to his rhythm and mood effortlessly. He imagines a platform where finding new music is as intuitive as his work, blending seamlessly into his routine.
Bio
Age: 28
Location: New York
Occupation: Data analyst
Hobbies: Playing guitar, hiking
Fav Genre: R&B, electronic
About Jonah
Who Have These Pain Points and Needs?
15
Frustrated
Adding songs to a playlist manually takes time.
Happy
It’s joyful to listen to the songs that matches my mood.
Disappointed
The songs in the playlist are repetitive and boring. I want to discover new songs.
Jonah’s Current Journey Map
Frustrated
Searching online for music takes time...
Goal
Listen to music that aligns with Jonah’s current mood
Expectation
Include new music (artists, songs, styles)
Instant way to listen to the music
Excited
I feel like listening to calm R&B music now.
FEELINGS AND THOUGHTS
ACTIONS
STAGES
Open Music Streaming App
Search for
Public Playlist
Play Public Playlist
Search for
New Music
Create Own Playlist
Play Own Playlist
Open music streaming app platform
Input current mood into search bar
Browse public playlists that the platform offers
Select a playlist
Play the selected playlist
Search for new music that aligns Jonah’s moods
Try listening songs
Check whether the songs align his mood
Add selected songs
Create a playlist on the platform
Play the playlist
Frustrated
It’s difficult to find a playlist that matches my current mood.
16
How Does Aria Compare to Competitors?
Marketplace for Aria by Amazon
Easy call back
Easy short prompt search
Discoverability
Easy playlist building
Language accessibility
Automation
Generative AI
Shortage of current music Apps
18
Competitive Highlight oriented Design Goals
Smart discovery through short prompts
Easy archive access
Automative playlist archive
19
Crazy 8s ideation session:
Individual ideation
Critiques and notes postings
Align with problem statement
Vote for directions
How Did We Ideate
the Aria Feature?
20
How Is Information Architecture Designed to Support Aria?
We did site mapping for Aria based on the 2 entry point. For users, it is more natural to add this feature to the process that they are already familiar with, which can help them discover and generate playlists faster.
21
Userflow
Navigate to Aria
Set music preference
Generate custom
playlist
Listen to
custom playlist
Set current music listening preference. User can customize music based on the following categories.
Moods
Genres
Ratio of new and familiar music
Custom preference
Start playing the playlist.
Open Amazon music app and navigate to Aria feature. Users can navigate to Aria from the following entry points.
Home
Library
Aria generates custom playlist based on users’ preference. Aria decides the following.
List of songs
Name of playlist
Cover image
Aria
User
User
User
23
Crafting Low-fi Wireframe
Homepage CTA entry
Generation setting
Country choices
Genre choices
Mood choices
Generated playlist
Edit playlist
While playing
Open playlist while playing
Check previous generated playlists
We draft the low-fi wireframe to visualize our idea and as a resources to show our users during testing.
24
Testing Low-fi
Can our design solve our customers’ problem?
Tested with 10 people
Moderated session
Findings
Wording of labels is confusing
Difficult to associate “Mode” with the ratio of familiar/unfamiliar songs, and “What’s more?” as the additional prompt input.
Moods options should be diverse and easy to select
Not only emotion influence users mood, so we need more diverse options. Images related to the mood help users understand what mood they are.
The tune setting options covers factors influences their music listening preference
Overall, users are satisfied with the tune setting categories. We also found the country is not necessary because users can specify country preference as an additional prompt.
25
High-fi Prototype - Entry Point
Introduce Aria as a newly AI powered feature cooperating with Amazon music APP. We present the follow two entry points:
Home page CTA
Capture awareness of new users by use of image and color.
Great visibility and accessible as on landing page.
Library
Easy archiving
Use as secondary entry point other than home page CTA
26
High-fi Prototype - Setting Preference
Moods
Genres
Setting main page
Match By
Addition
27
High-fi Prototype - Setting Preference
Moods
Users can select moods from the set of options including
emotions
time of the day
seasons
28
High-fi Prototype - Setting Preference
Genres
Users can select genre preference.
The top 8 genres shown on the tune setting page are determined by users’ past preference.
29
High-fi Prototype - Setting Preference
Match by
Users can customize the ratio of songs of current preference and new, unfamiliar songs.
In the low-fi testing, we learned “match by” is a new concept, therefore we provided the tooltip to explain it.
30
High-fi Prototype - Setting Preference
Addition
Users can add custom request for AI.
Example:
include Beyoncé songs
include 1990s songs
include Indian songs
31
High-fi Prototype - Playing AI playlist
AI playlist page
AI-generated image
Featured image is also generated by AI from the user’s custom prompts.
Playlist title
Playlist title is named by AI using the user’s custom prompts.
Tune setting prompt
The tune setting prompt is displayed to let users check after generating the playlist.
Gather insights:
User testing questions
Have you encountered any obstacles in completing tasks independently?
If yes, what is it?
If no, is this task easy for you to complete? Where do you think can still be clarified?
How helpful do you think this function is?
Users are given the example task in this session.
We observe the process they do the task and ask following questions:
Duration: 15min each Participant: 5
What We Can Improve
Add more detail interactions
Improve use of signifier
Careful on categorization
Users want to know how share, edit, like and other micro features work.
Use constant signifier for the “generation setting feature”. If a term is hard for users to understand, add an icon.
User feel “mood” is hard to categories and very personal. Maybe it’ll be better with have no categories.
What We Did Well
Before usability study
After usability study
Company
Password
Forgot Password?
4:30
Company
Password
4:30
Forgot Password?
Remember me
“I'm very busy at work so I can't spend much time discovering new music. I love the fact that the music list is infinite, providing plenty of room to explore. The options in the settings aren't complex either, and I was able to do it in a few minutes. Very simple and efficient. ”
“This feature puts the power of music discovery directly into listeners' hands, making it easier than ever to find and create personalized playlists. It's a game-changer for music enthusiasts, democratizing and personalizing the way we explore and enjoy new tunes.”
