Silver Interface, Interaction and User Experience Design · 2020
Musiac Music Recommendation Service
by Chia-Min Lin
- A' Design Award 2020 Silver Winner in Interface, Interaction and User Experience Design
- Awarded April 15, 2020
- CLIENT: LinStudio
- Estimated Number of Design Views: 55,707
- Verification code 73BFA4573B · Verify this award
About Musiac Music Recommendation Service
Musiac is a musical recommendation engine, utilize proactive participation to find precise options for its users. It aims to propose alternative interfaces to challenge the algorithm autocracy. Information filtering has become an inevitable searching approach. However, it creates echo chamber effects and constrains users in their comfort zone by strickly follow their preferences. Users become passive and stop questioning the options the machine provides. Spending time to review options may increase huge bio-cost, but it is the effort that creates a meaningful experience.
Press Kit
- Unique Properties
- Musiac is a music recommendation service aims to provide better selection mechanism. It uses a contour map to represent all musical genres in a single interface, showing the relationship across songs and genres in a 2D space. It maps out a taste bubble from the user's playlists, separating the user's prior musical taste field from the unexplored musical territory. The user can change the outline of the bubble by dragging it and expanding the scope of the recommendation algorithm.
- Dimensions & Packaging
- Musiac is a desktop application. It requires a user to import his/her playlists of third-party apps to build the "Taste Bubble" at the beginning. The bubble sits on an algorithmically-generated genres map, with a horizontal (electric, acoustic) and a vertical (atmospheric, bouncier) indication. The bubble separates the familiar and unexplored territory for a user so he/she could expand the recommendation (temporarily), without changing their behavior.
- Tags
- Algorithm autocracy, Ethical interface, Information architecture, Data visualization, User empowerment, Collaborative filtering
- Production Technology
- Musiac learns from "Every Noise at Once", has a virtual database, a theoretical possible algorithmically-generated map, and an ethical interface to reduce the bio-cost of finding liked songs from enormous options. It starts with a study of recommendation engines and filters, interviews, and a pre-testing, then ideates with "what if" questions along the generative design process, run through 2 round of user-testing to validate initial concepts and iterates. The project took a year to finalize.
- Design Challenge
- The most difficult part of this project is to find an industry professional to talk to me. Online streaming service is a very competitive field. Hence, business and advertising dominate the final result of music recommendations, which creates intransparent and unethical algorithms. Academics and scholars provide enthusiastic feedback, but one of the business executives cancel his help after he saw my questions list.
Another major research constrains is the fuzzy relationships of million users' playlists. Single users' taste bubble might not accommodate with the overarching dimensions, even though the machine could easily map his or her playlists into an individual lump. After I consulted with the famous founder of EveryNoise, he gives me great advice that I should have customized dimensions for different users to solve this issue. The system could map out the contour map from each individual users perspective, based on what's the location of the taste bubble, then decide how to show the rest of the genres. Despite the computation power is not efficient enough for this project, but this solution is definitely feasible at the nearby future. - Project Duration
- Musiac was a one year project done in 2018, Detroit.
- Operation Flow
- Musiac is a desktop application with an unusual interface, allows users to control the scope of recommendations and to manipulate the filtered results. It uses a bubble to separate the unexplored genres from the familiar genres. A user could easily change the shape of the bubble to expand the recommendation source. It also provides multiple filters, such as themes with different adjectives; numbers of views, finding songs with specific traits, and a guiding system to go beyond the comfort zone.
- Research
- Musiac ran with a waterfall and agile hybrid methodology: research, framing, prototype, test, and reflecting. Feedback was collected along the way and therefore, multiple iterations. I did solid case studies, interviewed active and passive music online streaming users, and consulted with programmers to finalize the proper frame of this project. Multiple user testings with a clickable prototype were performed in-person, as well as online.
- Inspiration
- Musiac started with a simple notion. Interface is the tangent point between our perception and reality. I believe interface reflects a subtle world of why we make choices. I ran into a website called Every Noise at Once and found it's a perfect database to test my theory. Interface has a higher-level goal of helping us to process information better. I want to design something that goes beyond data visualization and empower users to be more proactive upon the world they're seeing.
- Image Credits
- All materials required in this project doesn't need attributions. Licenses are attached in the documentation appendix.
Colors of Musiac Music Recommendation Service
Get Musiac Music Recommendation Service
Design Interview about Musiac Music Recommendation Service
The main idea is to map out all genres on the 2 dimension graphic and visualize user's taste in the shape of a bubble. User could drag the outline of the bubble to generate different recommendations. I referred to the philosophical definitions of a map and how it help us to find directions.
I've always been interested in how our brain process information. Human's cognition is a result of both internal and external. If interface represents a tangent point of our mind the real world, it shouldn't be considered as merely astheatic issue. So the goal behind this project aims to exploring what roles interface could play in the decision making process.
I am looking for opportunities to present this project. I've shown this project to data analysist and they've shown high interest. So maybe I will write a paper to describe the goal and intentions behind.
I spent probably 6 months to define what's the problem of searching and recommendation system before I landed on this mapping idea.
I've always struggled to tell Youtube or Google that "this is not what I'm looking for" because current systems don't provide feedback mechanism to their audiences. It's a really common frustration among its users, so, as an interaction designer, I tried to propose a different perspective of how to solve this.
I haven't think about it because this project is just a beginning. But I am happy to work with anyone who is interested in Musiac to make it into reality.
Lots of times when I'm listening to music, I feel I'm in a loop and couldn't find the way to expand my recommendations. I don't know how to break the loop for a long time until I started to see interface & experience design differently, then I see possibilities to propose the unconventional solutions.
Yes, I consulted this project with many people, my professor, my colleagues, the founder of EveryNoise, sociologist, and programmers. Lots of people have helped me along the design process.
Any person who struggled to expand their musical taste on current platforms. People who are tech-savvy, willing to spend time to do research to make sure they're not trapped in the comfort zone.
Musica has a very unusual interface, its UX was actually build on its UI. Musiac's interface determined how people see options. It's usually the opposite in current industry — build the framework of UX then UI polish the work. Musiac gives a further goal of data visualization to inspire users to proactively do something.
Musiac means " the maniac in music". Also has a slightly implication of the art of mosaic.
Paper/pencil. Adobe. Imaginations.
It nothing like current musical interface because it has a higher level goal of helping people process information better to fight against algorithm autocracy.
I consulted with many people along the way, but I build this project by myself so it really takes me a long time to finish it.
Algorithm is the key to map out all genres. There are over 40 millions songs exist on a single platforms, without the help of programing, it's impossible to visualize the fuzzy relations between each songs.
I did multiple contextualized research, such as interviews, remote user testing and behavior observation. The analytical research includes study reports on consumer behavior change, do questionnaire to seek people's feeling about songs, etc.
The business and advertising dominates the final results of recommendation. So the end users needs are actually contradicting with the need of industry. It might not have an immediate impact on my project since it's not on the market, but it hard to seek the right solution when there're other intention behind the algorithm.
My colleague told me I should try to publish my project so I start to see what are the competitions I could join.
I learned lots of history about algorithms and how it evolved to this complex level. The different goals of rating system and how it gradually become recommendation index. Also I learned lots of After Effects skills.