MAVL: Multiresolution Analysis of Voice Localization
summary
MAVL localizes a user with a single smart speaker and an eight-microphone uniform circular array by (1) separating coherent multipath AoAs via multi-resolution analysis (wavelet + frequency-bin differencing), (2) emitting acoustic probes and running an improved 3D MUSIC plus beamforming to recover room geometry, and (3) retracing paths to triangulate position. Implementation uses 5 cm and 9.6 cm radius arrays; reported median AoA errors are 1.49° and 3.33° for the top two paths, and median localization errors of 0.31m and 0.47m.
AI-written summary · about
chapters (7)
- 0:00Introduction and motivation
- 0:35Benefits of voice localization
- 1:50System challenges and design
- 3:52Multi-resolution analysis
- 7:36Room structure estimation
- 8:48Performance evaluation
- 11:25Conclusion
related talks
05mAcoustic Strength-based Motion Tracking
15mVILAM: Infrastructure-assisted 3D Visual Localization and Mapping for Autonomous Driving
21mTracko: Ad-hoc Mobile 3D Tracking Using Bluetooth Low Energy and Inaudible Signals ...
06mPeer-to-Peer Localization for Single-Antenna Devices
15mAcoustic Sensing and Communication Using Metasurface
22mUbicoustics: Plug-and-Play Acoustic Activity Recognition
06mUnlocking the Beamforming Potential of LoRa for Long-range Multi-target Respiration Sensing
18mVibrosight: Long-Range Vibrometry for Smart Environment Sensing