Content Creation
Redesigning Soundraw; And audio content generation software
Designing for Creative Ownership in Generated Music
Overview
Soundraw is an audio generation software used by content creators and musicians.
As part of a study of AI in creative workflows, our brief was to test the platform with it's target users and identify how AI was introduced into the creative workflow, and identify scope for better integration.
Approach
Understand how musicians work, and how they feel about generated audio content
Takeaway
Musicians did not enjoy using Soundraw, and were not open to incorporating it in their creation process
Feels like a stock library
4 out of 4 musicians compared Soundraw to a stock library.
The output is not upto professional standards
3 out of 4 musicians said they would not publish the content created.
Not enough control over what gets made
Why don’t musicians like Soundraw?
“Wouldn’t put it out as something I made …[I] didn’t create it, didn’t do anything.”
Lack of customisation options
4 out of 4 musicians did not find all the filter options without prompting
Underlying issue
Soundraw generated outcomes without enabling contribution, making the experience feel extractive rather than creative.
Following Sarah, a musician, in her creation process
without Soundraw.
Sarah conceptualises a unique base melody. She builds on her original creation with several more steps to produce the track.
Creation process with Soundraw
With Soundraw, Sarah generates a completed track with a click of a button. She has no control over the base melody, and does not feel any ownership over the end outcome.
Solution
Translating musician workflow insights into design principles
Through our research we determined that while creative workflows differed, every musician identified ownership over at least one stage of creation as essential. Our design principle was to give control over every step of the workflow, while allowing ideation and augmentation with generative AI.
Design
Translating the musician mental model to the interface
Musicians preferred interfaces that reflected the complexity and intentionality of professional music-making.
Musical instruments require extensive training and skill, and musicians did not enjoy using software that did not replicate these qualities.
The interface drew from established music-production environments such as Ableton and analog synthesiser hardware to surface control and edit capabilities.
AI was just another tool on software that looked familiar.
Design
Recreating the creative process through the user experience of the platform
The flow of the platform recreates musician workflow, allowing them to create a base melody, define the type of sound and then add on layers of sound.
Build a base melody
Adjust the sound type
Add and arrange the track
The flow begins by creating a base melody by playing it on a traditional digital interface, transcribing a melody created using analog methods or by using generative features as inspiration.
Build a base melody
Adjust the sound type
Add and arrange the track
Adjust the type of sound using a set of predefined 'types' or by accessing soundpacks defined by specific artist styles.
These sound packs offer a means of monetisation for musicians who wish to share their content.
Build a base melody
Adjust the sound type
Add and arrange the track
Edit and master the track using a layer based approach found of standard music editing software.
The platform makes recommendations around instruments and sound types based on genre.
Design
Design Principles for AI integration
The audio generation was positioned as one module of the larger creation process meant to act as inspiration or augment human created output. This manifested as recommendations for types of sound and base melodies to get the process started.
AI is incorporated in parts of the process, rather than replacing it.
The use of AI is optional and always has a manual alternative.
The artist is an active participant in the making process, giving them ownership over the final output.
AI Capabilities
Inspire me
The existing generative features are positioned as 'inspiration' rather than the end of the creation process.
Artist Touch
Musicians refer to their favourite artists for types of sound. 'Artist touch' allows well known musicians to monetise their 'Type of sound' by making sound pack available to incorporate in the generation process,
Recommended Sounds
At different steps in the process musicians can leverage recommendations regarding sound type, instruments used and other aspects of the audio development.
Results
We asked our research participants “Does this feel like something you made?”
Maybe? No. Maybe if I could write my own [music] notes?
Yes, but if the music everyone was making sounded the same as mine, then no.
Yes, especially from the ‘Type of sound’ and ‘Edit track’ view.
Design Reflection
Successful human–AI collaboration in creative tools depends less on automation capability and more on preserving a user’s sense of authorship.
For musicians, value emerged when AI supported creative exploration without displacing creative ownership.