A novel Time-Delay-of-Arrival estimation technique for multi-microphone audio processing

Multi-microphone speech enhancement requires knowledge of relative Time Delay of Arrival (TDOA) of the desired acoustic source at microphones. This paper presents a novel TDOA estimation method, Steered Null Error PHAse Transform (SNE-PHAT), which exploits null-steering to improve estimation robustness. The method is formulated to be computationally efficient.

A generalization to provide frequency-dependent TDOA estimates is proposed. Experimental results demonstrate that SNE-PHAT outperforms the Generalized Cross Correlation PHAse Transform (GCC-PHAT) method, particularly in the presence of background noise. Additionally, experiments illustrate the benefits of using frequency-dependent TDOA estimation.

Share This Post