Cura
CLIENT
Self-direct project
TIME
4 weeks
SKILL
UX Design, User Research, Prototyping
TOOL
Figma, After Effect, Premier Pro
MY ROLE
User researcher & UX designer
TEAM
Visual Designer
This project reimagines Spotify’s streaming experience.
We explored how empathetic intelligence and conversational user interfaces (CUI) could enhance one of the world’s leading streaming platforms, testing the concept through rapid prototyping and user feedback.
PROBLEM
Spotify is renowned for its personalized recommendations, using users’ listening history and their massive database to curate playlists like "Daily Mix" and "Discover Weekly" tailored to individual preferences.
However, as we delved deeper, we saw opportunities to enhance emotional support through music. Our emotion are often inconsistent, and users’ expectations of music as a source of emotional support vary greatly from person to person.
This leads us to consider:
How can Spotify evolve to truly support users through their emotion highs and lows in real life?
PROCESS
For this project, we engaged in multiple cycles of swift testing and iterations, with the goal of refining the concept through practical application.
1 /
Initial Research
User Interview
First, we explored 2 Spotify users' current music experiences, how they handle negative emotions, and their viewpoints on AI voice assistants.
Survey
We also conducted a quick survey to double confirm on the insights we gain from the interviews. With the 30 responses collected, we briefly examined the relationship between people's emotions and their music experiences, and we gained insights into how AI could contribute.
2 /
Iteration
User Scenario
We began by exploring the interactions between users and the AI assistant, listing five scenarios as examples and hooks to the features:
User discovers CURA on Spotify
Exhausted day
Late night
Bad weather
Sudden mood drop
Brand Identity
Our user testing revealed that colors play a crucial role in shaping how users perceive the experience. To align with this insight, we explored various mood boards that embodied the key traits of our concept: empathetic, calm, and personalized.
Ultimately, we chose the dark purple mood board and the name CURA to define our brand identity. The name resonates with the idea of “curated,” while the color purple, based on user feedback, evokes a sense of calm and aligns with our concept’s essence.
Concept Design
To better convey the key feature we plan to implement for CURA, we created a draft showing where this feature would be placed and the main content it might include.
3 /
User Testing
Scenario Test
We had two users evaluate our initial interaction concept. By creating scenarios and guiding them through the process through storytelling, we gathered valuable feedback on how users prefer the AI to react in different situations.
4 /
Iteration
Voice User Interface (VUI) States
Based on the updated user flow, we identified six states that CURA would display on screen while interacting with the user.
UI Prototype
Additionally, we developed a prototype featuring an updated UI and UX flow for the upcoming phase of usability testing.
5 /
User Testing
Usability Test
Based on the prototype, we conducted five usability tests to walk through the overall flow of the new feature. With the help of the usability test dashboard (as shown below), we received useful feedback about the UI screens, such as the placement of the buttons and how to make the content clearer, among other things.
6 /
Final Iteration
Finalize the Experience
We refined the screens based on user feedback above, and created a full prototype for the demo day. Overall, the UX flow remains unchanged, with only adjustments made to the content and UI elements.
7 /
Demo
Public Demo Day
During design week at Carnegie Mellon University, this project is showcased and presented to the public. Visitors were intrigued by the concept of curating personal playlists based on emotions, as it resonated with their listening habits. However, they also raised questions about the reliability of detecting emotions using wearable devices, suggesting that this could be a topic worth exploring further.
SOLUTION
We proposed
an AI-driven voice assistant integrated into Spotify’s streaming experience to foster feelings of intimacy, privacy, and understanding.
To accomplish this goal, we introduced CURA, the conversational assistant we are about to design.
As default, CURA lives in the existed interface of Spotify, and will automatically analyze your emotion through data collected from wearable devices and conversations* to tailor your very own experience.
(* note: technology inspired by hume.ai)
# CURATED PLAYLIST
Activate CURA just like how you call out Siri by saying "Hey CURA." You may ask it to play music based on your current mood.
# EMPATHETIC CHAT
Start a conversation with CURA if you need someone to talk with in those moments.
# RECAP
Visit the app to check out your recap—and revisit the songs that guided you out of the dark.
Designed to stand by your side whenever you need it, the more you interact with CURA, the better it will understand you.
Motion States
Based on a range of use case scenarios, we identified 6 distinct states that CURA would display on-screen while interacting with users.
Basic
The default state.Wanna Speak
Serving as a proactive agent, it will hint users when it has something to say.Speaking
When it starts to speaks.Listening
When it starts to receive inputs from users.Processing
When it starts to load data, analytics, or other information.Breathing
Only when users activate the meditation mode.
UI Prototype
Here’s a UI walkthrough from the user’s point of view as they are introduced to this new experience.
REFLECTION
Finding the right balance of proactiveness
Initially, we aimed to create a proactive agent capable of detecting when sadness occurs, prompting the user to engage by vibrating or making sounds. However, during user testing, we discovered that users were uncomfortable with this concept. One user expressed it well: "What if I'm in a car with friends, and CURA starts talking about my depression? I wouldn't want others to know my emotions." As brands increasingly integrate personal assistants into their services, it is worthwhile to conduct thorough research on how their target audience perceives building trust with this emerging technology.
User testing in more realistic environments
In this project, we employed rapid iteration, which limited us to reaching target users of similar age (mostly friends) through virtual interviews. Given more time, it would be beneficial to test the prototype in real-world settings and with a diverse user profile. For instance, a long-term diary study could track whether CURA effectively supports users through emotional fluctuations and provides a positive impact.
minjeanchu@gmail.com