Advanced

Robust Time Difference Estimation for Unknown Microphone Positions with Reverberation

Segerblom Rex, Simon LU (2015) In Master's Theses in Mathematical Sciences FMA820 20151
Mathematics (Faculty of Technology) and Numerical Analysis
Abstract
In this thesis a system is developed for robust time difference estimation for multiple microphones in reverberant environments, without any knowledge of the 3D positions of the microphones or the room. RANSAC is used to find line segments in sets of frames containing a number of the largest GCC-PHAT peaks. Focus is on real experiments with one moving sound source continuously playing music. The estimated time differences can be used to find the microphone positions, track the sound source, and estimate flat surfaces in the room. The system has been evaluated on real data with manually annotated ground truth in order to estimate the precision of the system in terms of true positives and false positives. The algorithms have also been... (More)
In this thesis a system is developed for robust time difference estimation for multiple microphones in reverberant environments, without any knowledge of the 3D positions of the microphones or the room. RANSAC is used to find line segments in sets of frames containing a number of the largest GCC-PHAT peaks. Focus is on real experiments with one moving sound source continuously playing music. The estimated time differences can be used to find the microphone positions, track the sound source, and estimate flat surfaces in the room. The system has been evaluated on real data with manually annotated ground truth in order to estimate the precision of the system in terms of true positives and false positives. The algorithms have also been incorporated into a system that takes a number of sound files as input and produces the microphone positions, sound path motion and reverberant structures. The complete system has been evaluated on real data with promising results. (Less)
Please use this url to cite or link to this publication:
author
Segerblom Rex, Simon LU
supervisor
organization
course
FMA820 20151
year
type
H2 - Master's Degree (Two Years)
subject
keywords
TDOA, RANSAC, GCC-PHAT, GCC, time difference, range difference, sound tracking
publication/series
Master's Theses in Mathematical Sciences
report number
LUTFMA-3271-2015
ISSN
1404-6342
other publication id
2015:E6
language
English
id
5276584
date added to LUP
2015-06-18 12:09:47
date last changed
2015-06-18 12:09:47
@misc{5276584,
  abstract     = {In this thesis a system is developed for robust time difference estimation for multiple microphones in reverberant environments, without any knowledge of the 3D positions of the microphones or the room. RANSAC is used to find line segments in sets of frames containing a number of the largest GCC-PHAT peaks. Focus is on real experiments with one moving sound source continuously playing music. The estimated time differences can be used to find the microphone positions, track the sound source, and estimate flat surfaces in the room. The system has been evaluated on real data with manually annotated ground truth in order to estimate the precision of the system in terms of true positives and false positives. The algorithms have also been incorporated into a system that takes a number of sound files as input and produces the microphone positions, sound path motion and reverberant structures. The complete system has been evaluated on real data with promising results.},
  author       = {Segerblom Rex, Simon},
  issn         = {1404-6342},
  keyword      = {TDOA,RANSAC,GCC-PHAT,GCC,time difference,range difference,sound tracking},
  language     = {eng},
  note         = {Student Paper},
  series       = {Master's Theses in Mathematical Sciences},
  title        = {Robust Time Difference Estimation for Unknown Microphone Positions with Reverberation},
  year         = {2015},
}