⌨️ The HEAR Dataset

A Comprehensive Acoustic Side-Channel Attack (ASCA) Corpus

Official dataset for the AsiaCCS '26 paper:
DECKER: Domain-invariant Embedding for Cross-Keyboard Extraction and Recognition

---

📖 Overview

Acoustic Side-Channel Attacks (ASCA) exploit typing acoustics to infer sensitive keystrokes. Historically, ASCA research has been constrained by datasets lacking diversity in users, environments, and devices.

We introduce HEAR, the most comprehensive and realistic ASCA dataset curated to date. HEAR is designed to evaluate keystroke inference algorithms along three critical axes:

1. Keyboard Generalization (cross-device robustness)
2. Noise Adaptation (environmental resilience)
3. User Bias (typing style variations)

✨ Dataset Highlights

• 👥 53 Participants: Spanning multiple genders, handedness profiles, and typing behaviors.
• 💻 37 Laptop Keyboards: Including built-in scissor-switches, external membrane boards, and mechanical switches.
• 🌍 3 Realistic Environments: Quiet (library), moderate noise (office), and high noise (café/outdoor).
• 🎙️ 3 Capture Modalities:
  1. External microphone capture
  2. Device microphone capture (no network noise)
  3. VoIP-based streaming capture (simulating real-world remote conferencing)
• 📝 Continuous Text Typing: Realistic sentence-level keystrokes, significantly harder to parse than isolated key presses.
• ⏱️ Multi-Recorder Synchronization: Explicit offset estimation enabling multi-channel, cross-device ASCA analysis.

🧠 About the DECKER Framework

This dataset serves as the benchmark for DECKER, our proposed 4-stage domain-invariant keystroke inference framework. DECKER utilizes:

1. Keyboard Signature Normalization (KSN)
2. Domain-Adversarial Disentanglement
3. Supervised Cross-Keyboard Contrastive Alignment
4. Acoustic Style Randomization (ASR)

📜 Citation

If you use the HEAR dataset or the DECKER framework in your research, please cite our AsiaCCS '26 paper:

@inproceedings{10.1145/3779208.3805984,
author = {Maurya, Bikrant Bikram Pratap and Choudhury, Nitin and Agarwal, Daksh and Buduru, Arun Balaji},
title = {DECKER: Domain-invariant Embedding for Cross-Keyboard Extraction and Recognition},
year = {2026},
isbn = {9798400723568},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3779208.3805984},
doi = {10.1145/3779208.3805984},
abstract = {Acoustic side-channel attacks (ASCA) on keyboards pose a significant security risk, as keystrokes can be inferred from the typing acoustics, revealing sensitive information on laptops. Prior studies on ASCA are limited due to dataset constraints, which include a smaller number of users, keyboards, and environments. This limits the exploration of the potential of this attack vector across different keyboards, users, microphones, and ambient noise conditions.To this end, we introduce HEAR, a novel dataset designed to study ASCA along three axes: keyboard generalization, noise adaptation, and user bias. HEAR contains recordings from 53 participants using 37 different laptop keyboards, collected in three realistic settings: (i) captured via an external microphone, and (ii) captured through the user's own device microphone over no network channel noise, and (iii) captured while in a network streaming platform over VoIP. This enables controlled benchmarks for cross-user, cross-keyboard, and cross-environment evaluations in a wider scope.On HEAR, we establish a new ASCA benchmark that spans conventional features and pre-trained representations from both raw audio and spectrograms, in unimodal and multimodal settings. We further propose DECKER, a domain-invariant keystroke inference framework that works in a four-stage generalization strategy: (i) Keyboard Signature Normalization to reduce device coloration in the waveform domain, (ii) domain-adversarial disentanglement to suppress keyboard identity, (iii) supervised cross-keyboard contrastive alignment to enforce key consistency across devices, and (iv) Acoustic Style Randomization to synthesize unseen keyboard responses. To explore the potential depth of the attack, we further investigate a sentence-level inference attack that leverages a large language model (LLM) layer as a post-processing layer, refining the predicted keystroke sequences using linguistic context.Empirical results on HEAR show that DECKER improves keystroke identification over strong unimodal and multimodal baselines, especially in cross-keyboard and cross-user settings, and that the language-model rectification further boosts sequence recovery. These findings highlight that ASCA remains effective under diverse users, devices, and noisy environments, underscoring its practical security risk.},
booktitle = {Proceedings of the ACM Asia Conference on Computer and Communications Security},
pages = {1707–1720},
numpages = {14},
keywords = {Acoustic Side Channel Attack, ASCA, Keystroke Inference, Deep Learning, Domain Invariance, Cross-Keyboard Generalization, ECAPA-TDNN, Contrastive Learning, Adversarial Representation Learning, Audio Processing, Security and Privacy},
location = {Bangalore, India},
series = {ASIA CCS '26}
}