On the Audio Representation of Radial Direction
Susumu Harada, Hironobu Takagi, Chieko Asakawa · 2011 · Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '11) · doi:10.1145/1978942.1979354
Summary
This paper proposes a new primitive for auditory displays — radial direction — and evaluates a concrete mapping of vowel-like sounds to eight cardinal and ordinal directions. The authors observe that almost all prior sonification work has mapped data to linear sound properties such as pitch, loudness, pan, or tempo, which suits scalar values such as coordinates but is awkward for inherently circular quantities such as headings, compass bearings, and the tangent directions that trace out a shape or path. Linear mappings wrap around discontinuously at the extremes, forcing the listener to perform a mental transformation, and two linear channels are needed to convey a single 2D vector. Drawing on the Vocal Joystick project, the authors observe that the International Phonetic Association vowel chart itself has a two-dimensional layout corresponding to tongue height and backness, so vowel sounds blend smoothly into a closed loop that naturally matches the topology of radial directions. They built a MIDI-based prototype that plays pre-recorded vowels (e.g., [i] for west, [a] for east, [æ] for upper-right) and ran a longitudinal study with five sighted and two blind participants across up to ten sessions covering three stages: sound mapping (learn direction-to-vowel associations), sound following (track a moving sonified direction with a joystick), and shape recognition (identify letters, digits, and free-form paths traced only in audio).
Key findings
All participants learned the four-direction mapping above 95% accuracy within a single session, with an average response time of 0.85 s. Five of the seven participants (including both blind participants) reached stable performance on the full eight-direction mapping within three to nine sessions, averaging 1.26 s per direction identification. In the sound-following task, all participants — including those who had not fully stabilised on the static mapping — could track continuous direction changes up to 2 Hz with over 90% accuracy and a response delay of about 0.77 s, indicating that smooth transitions between adjacent vowels were easier to perceive than isolated cues. In the shape-recognition stage participants successfully identified nearly every letter, digit, and random path presented to them after 66, 107, and 89 seconds of exploration on average; the fastest runs reached 13–26 seconds. Several participants reported synesthetic moments of 'seeing' or 'feeling' the shape from its vowel sequence, and commented that remembering the sound pattern was easier with vowels than it would have been with abstract tones.
Relevance
For accessibility practitioners designing non-visual interfaces — tactile-graphic exploration tools, blind-navigation aids, eyes-free mobile apps, or data-visualisation sonifications — this work offers a compact, learnable vocabulary for conveying 2D vectors with a single sound channel. It is particularly relevant to mobility systems that want to encode the direction of a point of interest, the orientation of a tangent along a path, or the heading of a user relative to a destination, all without tying up stereo/spatialised audio channels that the user may need for environmental awareness. Limitations include the small sample (n = 7, only two blind), the restriction to eight discrete directions rather than a continuously blended space, and the open question of how vowel sonification interacts with users whose native language shapes their vowel priors differently (several participants reported mismatches with Japanese, Spanish, or Dutch phonology). The authors explicitly frame vowels as one instance of a broader design space that could extend to other multidimensional sound properties.
Tags: sonification · auditory display · auditory interface · non-visual interaction · eyes-free interaction · navigation · blindness and low vision · assistive technology · human-computer interaction