Data Sonification for Screen-Reader Users: When and When Not to Use
Ather Sharif, Neha Aitharaju, Srihari Krishnaswamy, Jacob Wobbrock · 2025 · Proceedings of the 22nd International Web for All Conference (W4A) · doi:10.1145/3744257.3744263
Summary
This paper from the University of Washington assesses when data sonification — non-speech audio representations of data visualizations — is and is not beneficial for screen-reader users. Despite sonification dating back to 1974 and growing adoption in commercial products (Apple added it as a standard feature across all devices in 2021), its actual effectiveness for screen-reader users in real-world contexts like employment has remained unclear. The researchers conducted a need-finding survey of 106 screen-reader users (recruited through the National Federation of the Blind; mean age 49.8; 84 with complete blindness; 59 using JAWS, 25 VoiceOver, 17 NVDA) followed by in-depth semi-structured interviews with 12 screen-reader users, equally split between those with and without prior sonification familiarity. The study broadens the definition of "screen-reader users" beyond blind and low-vision people to include those using screen readers for temporary or situational needs such as motion sickness or light sensitivity.
Key findings
Only half of survey respondents found sonification at least "somewhat beneficial," but a critical familiarity gap emerged: 75% of users with prior sonification experience rated it "somewhat useful" or higher, versus only 27% of unfamiliar users — suggesting underutilization rather than ineffectiveness. 48% of respondents had never heard of sonification, and 63% had never encountered it in online data visualizations. Three themes emerged from qualitative analysis: (1) "Needle in a Haystack" — sonification is rarely encountered online, mostly only in research studies, and is not yet mainstream. (2) "Keep it Simple, Sonifier" — sonification is helpful for obtaining data overviews and trends ("the audio equivalent of a first glance") but has limited value for complex visualizations, multi-dimensional datasets, or granular data exploration. For low-cardinality data (few data points), participants preferred alt-text or tables instead. (3) "Suggestions to Improve Sonification's Effectiveness" — users want sonification combined with other modalities (alt-text, data tables), customizable parameters (pitch, frequency, speed), and explanatory metadata about what the sounds represent. The authors produced a decision tree for visualization creators: sonification is recommended when data has higher cardinality, portrays a clear trend, and communicates axis labels/values/ranges; it is not recommended for low-cardinality data (use alt-text/tables) or complex visualizations without complementary modalities.
Relevance
This is one of the largest empirical studies of screen-reader users' experiences with data sonification, and it arrives at a crucial moment: as data visualizations become ubiquitous in workplaces and on the web, their inaccessibility creates real barriers to employment and career advancement for screen-reader users. The paper's practical decision tree gives visualization creators an actionable framework rather than a blanket recommendation to "add sonification." For accessibility practitioners, the key takeaways are: sonification is best suited as an overview tool for trends, not for granular data extraction; it should be offered alongside other modalities (text descriptions, data tables) rather than as a standalone solution; and customization is essential because individual hearing perception varies widely. The finding that familiarity dramatically improves perception of sonification's usefulness suggests that the technology's limited adoption is partly a discoverability and education problem, not just a design problem. The "one size does not fit all" conclusion — with users requesting customizable pitch, speed, and frequency controls — echoes findings across accessibility research that personalization is not optional.
Tags: sonification · data visualization · screen readers · blindness · visual impairment · audio graphs · web accessibility
Standards referenced: WCAG 2.1