University of Michigan · Engineering Interactive Systems course project · Jan–Apr 2023

Showing sound direction with LEDs

SoundAround: Wearable Sound Localization

The original SoundAround helmet prototype with microphone array, Raspberry Pi, battery pack, and LED wiring, positioned in the five-direction study setup.
The course prototype during direction-identification testing. The five numbered positions marked sound sources in front of the wearer.Original course report, Figure 3 · Le Qin and Jeremy Zhengqi Huang

SoundAround is a wearable prototype that uses LEDs at the edge of the wearer's vision to show where a sound is coming from. A microphone array estimates the direction, and Python software chooses which lights to turn on.

My role

I came up with the idea and worked on hardware assembly, Python software, interaction design, testing, and results analysis. I didn't lead the participant study sessions.

participants in the final evaluation
5

Quiet-lab tasks with noise-cancelling headphones

front-direction identification
88%

22 correct responses across 25 trials

rear-direction identification
93.3%

14 of 15 trials; approximate pointing counted

Starting point

Keeping the cues in view

I proposed SoundAround with deaf and hard-of-hearing people in mind. Placing the lights in peripheral vision would let someone check a sound's direction while still looking at their surroundings.

How it came together

  1. 01

    Read the microphone array

    A ReSpeaker four-microphone array sends direction estimates to a Python process on a Raspberry Pi.

  2. 02

    Choose the lights

    A separate display process reads those estimates and lights the corresponding LED positions. Red and green distinguish sounds in front from sounds behind the wearer.

  3. 03

    Adjust the display

    Early feedback led to changes in how directions mapped to LEDs. The revised display used lower brightness, added diffusion, and dimmer idle lights to make the cues easier to read and the device more comfortable to wear.

System diagram connecting a four-microphone array, Raspberry Pi 4, programmable LED strip, and portable power supply.
The microphone array sends direction readings to the Raspberry Pi, which controls the LED strip. A portable battery powers the prototype.Original course report, Figure 1
Original interface sketch showing light position and color indicating front and rear sound directions around a wearer.
The original sketch shows how LED position and color represent sound direction.Original course report, Figure 2

Background noise made the lights harder to read

Participants could read the cues in the quiet lab. Outdoors, background noise could make the lights unstable and increase the time needed to interpret them. Further work would need to address noise filtering and test the display in everyday settings.

The percentages describe how often participants identified the tested direction. They don't measure the microphone's angular accuracy. Rear trials counted pointing at or very near the target as correct.

Participants wore noise-cancelling headphones to limit what they could hear. This doesn't reproduce the experience of being deaf or hard of hearing, or establish how well the device would work for those users.

The source archive contains experimental hardware and software variants with known defects. It records the course prototype and hasn't been verified as a reproducible release.

Reports, code, and credits

Course reportHistorical source ZIPCode on GitHub

I worked with Jeremy Zhengqi Huang, and we wrote the course report together.

The ReSpeaker device supplied the direction estimates. My work connected those readings to the wearable display and examined how people interpreted it.

Next projectChanging one feature in a generated face