BudsID: Mobile-Ready and Expressive Finger Identification Input for Earbuds

TL;DR

BudsID demonstrates that finger identification using a finger-worn magnetic ring and an earbud magnetometer is a viable, expressive input modality for wireless earbuds. Across two user studies, single-tap finger classification achieves up to 96.9% accuracy with fast interaction times, while a novel multi-finger double-tap scheme reaches up to 94.7% accuracy, maintaining low error rates in mobile scenarios. The work provides an open-source hardware/software platform and assesses practical application designs, including contextual and hierarchical input, as well as a user feedback study showing favorable usability. Overall, finger identification on earbuds can significantly enhance eyes-free interaction on mobile wearables, with strong implications for multi-tasking and IoT control in everyday contexts.

Abstract

Wireless earbuds are an appealing platform for wearable computing on-the-go. However, their small size and out-of-view location mean they support limited different inputs. We propose finger identification input on earbuds as a novel technique to resolve these problems. This technique involves associating touches by different fingers with different responses. To enable it on earbuds, we adapted prior work on smartwatches to develop a wireless earbud featuring a magnetometer that detects fields from a magnetic ring. A first study reveals participants achieve rapid, precise earbud touches with different fingers, even while mobile (time: 0.98s, errors: 5.6%). Furthermore, touching fingers can be accurately classified (96.9%). A second study shows strong performance with a more expressive technique involving multi-finger double-taps (inter-touch time: 0.39s, errors: 2.8%) while maintaining high accuracy (94.7%). We close by exploring and evaluating the design of earbud finger identification applications and demonstrating the feasibility of our system on low-resource devices.

Figures (8)

• Figure 1: BudsID prototype and sensing direction: plan views of disassembled (A) and assembled (B) earbuds. When a user wears BudsID (C), a magnetometer in an earbud-mounted microcontroller measures tri-axial variations in the surrounding magnetic field (D).
• Figure 2: Sample magnetometer data captured in the five different potentially noisy environments. 1. Static: static sitting posture, 2. Playing Music: wearing BudsID mounted on a commercial earbud playing music at a default (mid-range, comfortable) volume, 3. In the car: seated in a car, a potential source of magnetic interference, 4. Walking: walking in a straight line, and 5. Rotating: rotating clockwise in place.
• Figure 3: A) The mean normalized magnetometer data captured during single taps by five different fingers with two different sizes of magnets in the pilot study (N=6) and B) mean comfort ratings for taps by each finger
• Figure 4: Confusion matrices (% accuracy for general model and mean % accuracy for individual and LOOCV models) for deep-learning (A) and machine-learning (B) classifiers.
• Figure 5: Interaction plots of inter-touch time (A), error rates (B), and self-reported usability (measured with 5-point Likert Scale) (C) for different sequential taps of first and second fingers.