Core Concepts
The author proposes a novel method utilizing hybrid TDOA information to calibrate asynchronous microphone arrays, demonstrating independence of microphone number and improved calibration accuracy.
Abstract
The content discusses the calibration of asynchronous microphone arrays using hybrid TDOA information. The proposed method combines TDOA-S with TDOA-M and odometry measurements to achieve better calibration accuracy and stability. The study includes simulations and real-world experiments to validate the effectiveness of the method. Key points include the derivation and extraction of TDOA-S, the use of Graph-SLAM for calibration, and the application of the Gauss-Newton method for solving nonlinear least squares problems. Results show that the proposed method outperforms existing approaches in terms of independence from microphone number, insensitivity to initial values, accuracy under various TDOA noises, and lower CRLB for microphone parameters.
Stats
Simulation results consistently prove that our method is independent of the number of microphones.
Our method has better accuracy and stability in estimating microphone parameters under various TDOA noises.
Real-world experiment results confirm that our method performs independently of the number of microphones.
Quotes
"Our main contributions are stated as follows."
"A novel and efficient measurement: TDOA-S without time offset is proposed."
"Our method is independent of the number of microphones."