In the art of transforming data into easily interpretable displays, visualization is usually the go-to option. But sight isn’t the only sense we can use to make sense of numeric information.
Sonification is visualization’s lesser-known cousin: the practice of translating data into sounds. While we may not think of it explicitly, sonification is common in everyday life. For example, a car’s parking assist system takes the distance between a car and an object in its path and transforms it into a beeping frequency, increasing a sense of urgency in the driver as danger gets closer.
In the car alert example, the main benefit of using sound rather than a visual display is that the driver will receive the urgent information regardless of which direction they are facing. This is a great use case for sonification: It can free a doctor to focus visually on a delicate surgical operation while tracking a patient’s vitals via auditory heart monitor cues, or it can provide real-time peripheral awareness for operators in busy air traffic control rooms. However, there are also cases where the human ear is able to pick up patterns in data that the eye would simply miss.
Stephen Roddy believes sounds will be increasingly important for interpreting complex datasets. Joao Lima
Stephen Roddy, a researcher and lecturer at University College Cork, has spent much of his career transforming data into sound and developing novel sonification techniques and use cases. Perhaps unsurprisingly, he is also an active musician, often interleaving his technical and artistic pursuits. Roddy believes that sonification is primed for a resurgence, as a means of making sense of a deluge of data in a variety of fields. He is especially optimistic about sonifying networking data in the age of 6G, and is contributing to the development of an emerging field researchers call the “Internet of Sounds.” We caught up with Roddy to understand why this may be a coming trend.
Sonification’s benefits, then and now
IEEE Spectrum: In what situations is sonification useful?
Stephen Roddy: Sonification is primarily useful when a user’s visual attention must be directed elsewhere, or when traditional data graphing and visualization methods completely break down.
However, sonification truly excels when dealing with highly complex, multidimensional, and noisy data where crucial patterns and relations are invisible to the naked eye. While the human eye operates like a serial processor—focusing on one small chunk of information at a time—the human ear acts as a parallel processor. Two centuries of psychoacoustics research have found that the ear can effortlessly identify minute variations, process multiple data streams simultaneously, and isolate important signals hidden deep within background noise.
How new is sonification?
Roddy: It feels like a modern concept, and it’s a modern term. But sonification has actually been around for much of the history of technology. One early example is church bells in Middle Ages Europe. Distinct ringing patterns provided timekeeping information, markers of religious ceremonies and community events, and warnings of approaching danger.
Then in the mid-19th century, telegraph registers [printers that receive incoming Morse code messages] started out by recording received “dots” and “dashes” as indentations on moving paper tapes. Reading the printed data was slow and challenging work, so skilled operators quickly learned to identify the distinctive rhythmic clicking sounds produced by the registers. By the 1850s, the telegraph registers were replaced with sounders. These devices amplified the rhythmic clicks into “dits” and “dahs” and allowed operators to process the data more quickly and efficiently than reading dots and dashes on paper.
Sonification continued to anchor early telecommunications breakthroughs until the 1920s, when the telephone and teleprinter were commercialized. At that point, sonification and the broader field of communication technologies began to temporarily diverge.
Why is sonification having a resurgence now?
Roddy: In recent years, the sheer volume of data has skyrocketed. It’s becoming far too massive for visual interfaces to handle alone. By 2026, the global datasphere is expected to reach a staggering 221 zettabytes [1021 bytes]. Fields like space science manage datasets in the 20-to-50 exabyte [1018 bytes] range, and biological sciences handle up to 200 exabytes of data, with genomics data alone growing up to 40 times as fast as space data.
As these massive datasets grow, they become highly complex, heterogeneous, and filled with variables that interact in non-trivial ways. Crucial signals can be buried across wildly different scales, from near-instantaneous fluctuations to changes over long eras. Traditional charts can’t put all of that disparate data together in a sensible way, so the data becomes siloed. Researchers are turning to sonification as a necessary tool to break through this data deluge and extract meaningful insights.
What about AI? Can the processing and pattern finding be done by machines?
Roddy: Yes, AI can do a lot. AI techniques can organize, filter, and analyze a data set. But sonification can provide support to a human decision maker, by making patterns in the data perceptible to a listener acting on that data.
Recent research shows that AI-assisted sonification can be especially beneficial, with AI pre-processing and screening data. Then sonification presents the final product to a user in an easily interpretable way. For example, a team used AI to translate electroencephalography (EEG) data into sound to make it easier to interpret.
Modern sonification techniques
What are the most common techniques for mapping complex data to sound?
Roddy: Two important techniques used in the field are audification and Parameter-Mapping Sonification (PMSon).
Audification starts with low-frequency signals and transposes them into the range a human can hear easily [20 hertz to 20 kilohertz]. Often this just speeds up the track. This process inherently compresses the time-series data. It’s very useful for spotting anomalies.
A great example is gravitational waves. In 2015, LIGO scientists converted gravitational-wave data into the audible spectrum, and you could literally hear the collision of two black holes that were billions of light-years away. It sounded like a distinct “chirp” rising above the loud background noise. Also, Robert Alexander, a research scientist and composer, took solar wind data from NASA and sped it up from hours into seconds, and he could hear this peculiar harmonic hum in the carbon ion data that statistical visual models had missed. Turned out it was a particular behavior of the Sun’s magnetic fields.
PMSon is a bit more complicated. It maps specific variables or dimensions of a dataset onto distinct auditory parameters, such as pitch, volume, or instrument type. This works better for discrete, non-time-series data that contains diverse or mismatched data types.
For PMSon, a good example is the “Brain Stethoscope” project which converted 18 channels of an electroencephalogram (EEG) into sound. The mapping was designed so that normal brain waves would become this quiet background hum, but silent seizures would come out as an intense screaming sound. Untrained medical students would listen to this and accurately detect seizures 95 percent of the time. With normal visual scans, they could only identify seizures 50 percent of the time.
Another cool example was in genetics. A researcher named Mark Temple used PMSon to map the four repeating characters of DNA onto musical components like melody, harmony, and percussion, so that listeners could easily hear genetic mutations and biological motifs that would otherwise look like a massive, unreadable wall of text.
How do you use sonification in your own research?
Roddy: In one recent project, I used PMSon to help make sense of the visual clutter on network dashboards. I designed this for a national Low-Power Wide-Area Network (LPWAN) testbed that tracked network health across 10 different data streams simultaneously throughout Ireland. The idea was to synthesize clear, pleasant, voice-like sounds when the network was healthy, and shift into harsh, dissonant sounds when there are technical issues. You’re combining a bit of music, data analysis, and cognitive science here to make it easy for a human to just sit there and hear when there’s an issue.
Your browser does not support the audio tag.
In another project I sonified some smart city data in Dublin. The goal was to help citizens and administrators make decisions. I used a machine learning model called a Variational Autoencoder that helped take Dublin’s chaotic transport, traffic, and weather data and transform it into these pretty musical passages.
Your browser does not support the audio tag.
What do you think the role of sonification will be in the coming years, especially upon the arrival of 6G?
Roddy: As we approach the 6G era in the mid-2030s, I’m expecting communication to transition toward the “Internet of Sounds“ (IoS). Networks will soon transmit such massive amounts of data throughputs with such low latency that new things will become possible. You could have, for example, real-time, synchronized performances from 100-piece symphony orchestras, where the players are actually all over the world. Or, massive wireless acoustic sensor networks spanning thousands of kilometers, and life-size 3D holograms with spatialized audio.
These networks will be so complex that sonification will basically become necessary to make any sense of what’s going on. It’s funny, sonification and communication kind of split apart in the 1920s after the telegraph, but now the two fields are coming back together. But this time we won’t just be using sound to send messages over a network, we’ll be using sound as the primary tool to diagnose, navigate, and understand the massive, living data networks we have built.
From Your Site Articles
Related Articles Around the Web
Be the first to comment