Autonomous Learning of Speaker Identity and WiFi Geofence From Noisy Sensor Data

Chris Xiaoxuan Lu, Yuanbo Xiangli, Peijun Zhao, Changhao Chen, Niki Trigoni, Andrew Markham

Research output: Contribution to journalArticlepeer-review

Abstract

A fundamental building block toward intelligent environments is the ability to understand who is present in a certain area. A ubiquitous way of detecting this is to exploit unique vocal characteristics as people interact with one another in common spaces. However, manually enrolling users into a biometric database is time-consuming and not robust to vocal deviations over time. Instead, consider audio features sampled during a meeting, yielding a noisy set of possible voiceprints. With a number of meetings and knowledge of participation, e.g., sniffed wireless media access control (MAC) addresses, can we learn to associate a specific identity with a particular voiceprint? To address this problem, this paper advocates an Internet of Things (IoT) solution and proposes to use co-located WiFi as supervisory weak labels to automatically bootstrap the labeling process. In particular, a novel cross-modality labeling algorithm is proposed that jointly optimizes the clustering and association process, which solves the inherent mismatching issues arising from heterogeneous sensor data. At the same time, we further propose to reuse the labeled data to iteratively update wireless geofence models and curate device specific thresholds. The extensive experimental results from two different scenarios demonstrate that our proposed method is able to achieve twofold improvement in labeling compared with conventional methods and can achieve reliable speaker recognition in the wild.
Original languageEnglish
Pages (from-to)8284-8295
Number of pages12
JournalIEEE Internet of Things Journal
Volume6
Issue number5
Early online date3 Jul 2019
DOIs
Publication statusPublished - 1 Oct 2019

Keywords

  • access protocols
  • computer network security
  • feature extraction
  • Internet of Things
  • learning (artificial intelligence)
  • speaker recognition
  • wireless LAN
  • association process
  • inherent mismatching issues
  • heterogeneous sensor data
  • reliable speaker recognition
  • autonomous learning
  • speaker identity
  • WiFi geofence
  • noisy sensor data
  • intelligent environments
  • unique vocal characteristics
  • biometric database
  • vocal deviations
  • audio features
  • sniffed wireless media access control addresses
  • MAC
  • specific identity
  • particular voiceprint
  • supervisory weak labels
  • labeling process
  • novel cross-modality labeling algorithm
  • clustering
  • wireless geofence model
  • Wireless fidelity
  • Labeling
  • Speaker recognition
  • Noise measurement
  • Feature extraction
  • Wireless sensor networks
  • Cross-modal association
  • Internet of Things (IoT)
  • speaker identification

Fingerprint

Dive into the research topics of 'Autonomous Learning of Speaker Identity and WiFi Geofence From Noisy Sensor Data'. Together they form a unique fingerprint.

Cite this