AI (Artificial Intelligence) has now become integrated in our daily lives. People get help in writing text, brainstorming, and even drawings. What about music?
SoundMIX helps indie game developers, who lack assets and resource, to easily get proper sound and game music they need, delivering their exact intentions. This project dives into delivering the efficient way to put in diverse prompts in order to generate proper sounds that users desire. SoundMIX also lets users to check copyright as well as shouting out to original artists when needed.
INTRODUCTION
ROLE
UXUI Design, UX Research, A/B test, Final design
DURATION
4 months : 2023.03 ~ 2023.06
PROJECT SUMMARY
THE PROBLEM!
“Indie game developers need an easy and quick way to find royalty free music or sound effects for their game which delivers the game’s context properly, thus helping them to improve efficiency of their production pipeline ”
First, what I dealt with this project -
Diverse inputs
3 design targets
Editing
Consider copyright issues
Final Design Prototypes
SOLUTION
Flow 01 prototype : Users choose music or sound effect for the AI to generate
Users can put either text, short sound files, image files, or all of them to tell the AI the mood they desire
Users can type exact detail of music (beat, instrument) or describe a storyline they want
Generating music or sound effect
Flow 02 prototype : 
After AI sound generation, users can edit them manually to add their exact preference
Users are able to see history of their generation in case they like the first draft better
Editing manually after AI generation
Flow 03 prototype : Checking copyright and highlighting original artists if they agreed
Users must check copyright issues before saving
Original artists who have agreed to use their music as reference are highlighted and promoted
After editing or deleting the violating parts, users are able to save their sound files
Let’s see the design process!
INTRODUCTION
Project Context
We are undergoing dynamic change with various AI products emerging in market. There has been AI already, however, in the past few months with ChatGPT being so popular, people are talking about AI everywhere and the usage of them. There are AI for research, drawings, art, and in many other divisions.
What about in music industry? Music is another essential part of art that many need in production.
Can AI help people who do not have enough asset but who needs good music for their production?
RESEARCH
Netnography
Almost all of them in this Indie developer community are working on their own or in a small team (85%).
However, most indie developers have no budget for their production (69%).
Although there are massive corporation produced games and AAA games, the Indie games account for over 95% of all games on Steam
“Game, Music, Indie Game Developers”
RESEARCH
Survey & Interviews
Over 85% of participants answered that music generating AI improves productivity for music making people.
Features that users think is necessary
2. People considered features that let users type diverse prompt types and checking copyright issues necessary for music generating AI products.
Music is considered an essential element to deliver a game’s context since it connects with emotion
There are 3 ways to get music, but many prefer to make their one since it cannot deliver 100 percent their intentions
People from the survey and interviews considered functions that enable users to put diverse inputs and checking copyright the most essential feature needed in music generating AI products.
Takeaways - about indie game devs
RESEARCH
Industry Comparative Research
There are 4 main frustrations user face : time, edit, copyright, and non-human like sounds
Some users spend too much time finding proper music or waste their time trying to change the AI generated music
Many users consider editing and checking copyright issues as necessary features of Music generating AI
Non-human like sound can be solved by adding some human touch, such as editing
Takeaways
DEFINE
The Problem!
Target Audience
Who works alone or in a small team with less assets, considering music or sound effects an essential part for their game
Who cannot find the exact sound that delivers all of their intentions
“Indie” Game Developers
“Indie game developers need an easy and quick way to find royalty free music or sound effects for their game which delivers the game’s context properly, thus helping them to improve efficiency of their production pipeline ”
The Problem is …
Diverse inputs
Editing
Consider copyright issues
DEFINE
2 types of personas
01 The primary user is Nate, who makes games by oneself. Nate aims to make a unique game music that conveys his intentions 100 percent. He wants to reduce his time and effort when finding music.
01 The secondary user is Kelly, who is a illustrator looking for music that fits well with her personal contents. Kelly finds it difficult to right text prompts in AI products.
USER TEST
A/B Test with wireframes
“Users wanted to put diverse prompts to generate AI easily and edit them after manually”
Concept A :
Single page that user can put in text, sound files, and img files
Efficient and fast putting in diverse input sources at once before generating AI.
Users can see the overall sources they added at one sight which makes it easy to edit afterwards.
prosCan look a little cluttered when there are a lot of sources added.
When there are several sources, the page will extend long which makes users to scroll a lot.
cons
Concept B :
3 Separate pages that user can each put in text, sound files, and img files
Users can filter and see their 3 types of input sources. (text, sound file, image file)
This way of displaying is visually organized and easy to edit in each category.
prosText, sound files, and image files each have a separate page which makes users to click several times.
This way is efficient when there are a lot of sources, but when there are only few inputs it is not necessary.
consUSER TEST
A/B Test Results
DESIGN
Styleguide
DESIGN
Final Design 01
Users choose music or sound effect for the AI to generate
Users can put either text, short sound files, image files, or all of them to tell the AI the mood they desire
Users can type exact detail of music (beat, instrument) or describe a storyline they want
Generating music or sound effect
Flow 01 prototype : flow 01 description -
Start Page
Users can try SoundMix for free to generate royalty-free music or sound effects with the help of AI
Emphasize CTA button to users so that they can try for free quickly
Display what this website is for, to generate game music and sound with help of AI
WHAT I DID & WHY
Sound Generation Page
Users choose if they want ‘Music’ of ‘Sound Effect’ by clicking/tabbing the top buttons
Users can upload text prompts, sound files, or image files to deliver what kind of vibe they want their sound to be
Upload files in one page so that users can see their inputs at once
Enable users to drag and drop files from their devices
Highlight ‘AI Generate’ button
WHAT I DID & WHY
Sound Generation Page
Users can write only texts or choose to upload sound or image files when they feel they need it
See the ‘name’ of sound or image files
Users can either describe about the sound not only writing about which beat and instruments it will use but also stories they want the song to deliver
Since before generating, users will remember what files they are uploading so only the names are shown
WHAT I DID & WHY
Animation while users are waiting
Loading …
DESIGN
Final Design 02
After AI sound generation, users can edit them manually to add their exact preference
Users are able to see history of their generation in case they like the first draft better
Editing manually after AI generation
Flow 02 prototype : flow 02 description -
Generation Result page
Users can see their prompts or sources they put in one sight and regenerate them when needed
Designed to be a single page for users to put diverse sources after user testing which makes them easier to add or delete their sources
Deleted ‘view more’ from the wireframes and expanded the page so that users don’t have to click one more time to see their inputs
‘Edit’ button for users to edit manually
WHAT I DID & WHY
Edit Selection page
Users are able to make the AI regenerate for certain parts by clicking EDIT button and putting separate prompt for that selected part
The function to edit selected parts from the generated sound is to give a more human touch
A lot of users thought AI generated sounds sounded like it was made by AI which was not helpful for them to use
WHAT I DID & WHY
2nd Generation Result page
After regeneration, users can see the history of their 1st draft and can also edit the sources and regenerate again
Display the latest generated sound file at the very top, then editing section, then past drafts (History)
This is to make users edit/regenerate faster
Displaying ‘History’ of generated sounds is to let the users see and listen to them before saving
WHAT I DID & WHY
DESIGN
Final Design 03
Checking copyright and highlighting original artists if they agreed
Flow 03 prototype : Users must check copyright issues before saving
Original artists who have agreed to use their music as reference are highlighted and promoted
After editing or deleting the violating parts, users are able to save their sound files
flow 03 description -
Generation Result Page
Copyright issue is the most controversial part when using AI to generate music
Stress ‘Copyright Check’ button with accent color
Users must check copyright before they save the sound files
WHAT I DID & WHY
Checking Copyright …
Animation while users are waiting
Copyright Check Page
Users can see which parts violates copyright and which parts are referenced by original artists who agreed to let people use their music
Edit or Delete the violated parts which are displayed red
Shoutout to orignial artists who agreed to let people use their songs
For artists who made a contract with SoundMIX are promoted by adding tags and links to their album or websites
WHAT I DID & WHY
After checking copyright
‘Check Copyright’ button is changed to ‘Save’ button
Added check icon next to copyright icon letting the users why they now can save files
Users can now save this file to their devices
WHAT I DID & WHY
WRAP UP
Takeaways & Future Exploration
AI “helping” people who does not have enough asset & resource
There are ongoing debate about the usage of AI. AI has been implemented in many products we use actively in the past few months and is helping people to work fast and efficiently. Although there are some concerns about certain jobs taken away, AI can be handy when it is used in the right way ‘for’ human.
I want to emphasize that this project is to “HELP” people, indie game developers, who lack asset in their production. With the help of AI, they can save time and money in finding proper music for their game.
There were mainly 2 frustrations that users faced: finding a precise and easy way to generate AI, and copyright issues of the music made by AI.
I will pursue my research on the copyright issue part. I hope to discover a strong way to shout out for the original artists so that people using the AI music know that some music might be referenced by existing artists.
Further exploration
There were only few people for A/B test and high fidelity user test. If I have a chance, I hope to test enough people to drive more accurate results, and sympathize with diverse perspectives that I cannot anticipate.
Data driven results
