Acoustic Garden

As the first branch of Acoustic Garden Program, this project explores accessibility and affective computing through spatial audio in augmented reality. When the virtual world is mostly invisible, sound becomes the core of narrative and guides us to rediscover our surroundings. As we listen, and discover, our movement distorts the soundscape in turn, constructing an spatial audio narrative. Finally, as we find the sound, a virtual garden “bloom” above both the visual and acoustic world.



USC Grad Thesis

Built with Unreal Engine/MetaSound

Artist/Director – Yufan Xie

Instructor – Lisa Little

A human perspective upon urban space through sound

In the world of garden, bees do not rely on vision to locate flowers, but guided by pheromones carried in the air. They glide through the garden, alighting gracefully on each blooming flower. This process symbolizes a deeper mode of perception, resonate with today’s augmented human perception, enabling alternative ways to sense and understand the world. 


Spatial Music Innovation

In AR/XR design, spatial audio and audio-driven narratives are generally considered as secondary roles. Non-visual content production and research for visually impaired users are underrepresented. Similarly, in the field of music, visual interfaces dominate discussions on gesture-based sound synthesis, leaving intuitive, audience-driven experiences unexplored. Historically, studies in architecture and music focused on spatial and musical sequences, but they largely remained confined to print media representations. 

In this project, inspired by the progressive structure of electronic music, we introduce a spatialized sound synthesis method based on distance-related audio effect modulation combined with binaural spatialization. This approach is intended to navigate users through space without depending on visual indicators, utilizing auditory cues from a multitude of virtual audio objects instead. These objects are responsive to user movements, providing immersive musical experiences on a walking scale. Interaction with specific audio objects enables users to dictate different musical progressions, which follows a self-similar spatial structure. 

Unlike traditional sonification methods that mainly replicate or translate data, our approach emphasizes the emotional impact of auditory messages. It unveils the potential for audience-involved, spatially-driven musical narratives. Testing across various platforms, we faced challenges in sound design, hardware limitations, and cognitive processing.





Prototype Test on Screen


  1. Spatial data as a wavetable
  2. Disable volume attenuation, use distance related modulation to inform relationship to audio objects
  1. Users rely on visual interface and visualized target,

  2. Amplitude attenuation for spatialized audio create excepted silent area for user, and users get disconnected


  1. Audible range beyond visible range
  2. Disable volume attenuation, use distance related modulation to inform relationship to audio objects
  3. Using Bone Conduct/non-immersive headphone to avoid potential danger for blind person
  1. User affected data duplicate with their distance to object and conflict with absolute data distribution
  2. Sometimes user lose tracking during the silent gaps between audio cues
  3. Repetitive pattern reflects non-incremental variations, confuse user and disconnect with their movements


  1. Removing repetitive pattern along X,Y,Z directions
  2. Direct distance to Audio Object for fx modulation
  3. Simple audio design for accessibility
  1. Audio Spatialization is still delayed 
  2. Message overload and readability
  3. Readability vs experience Accessibility vs Affective


Accessibility, and spatial experience via interactivesounds


Navigation Test with Handheld AR


Visual Test on Street


Multi-Sequential & Multi Chapter Soun Narrative Test



Multi-Sequential Test for Musical Narrative


Multi-Sequential Test for Musical Narrative


Spatial Data Sonification Test