Multifractal analysis of visualized room impulse response for detecting early reflections.

This paper describes an improved method for detecting early reflections in the initial part of the room impulse response using multifractals. The proposed method uses the two-dimensional multifractal analysis. The room impulse response is visualized as a spectrogram image which is then subjected to the multifractal analysis. The algorithm is based on describing local regularity in the image using distribution of Hölder exponents. The time positions of the selected Hölder exponents in the image are utilized in detecting early reflections. The obtained results show better efficiency of the proposed algorithm compared to the previous one-dimensional multifractal analysis based algorithm.

Detection of early reflections using multifractals.

Early reflections are not always strong and sparse enough on the timeline to be easily identified when reviewing the impulse response diagram. This paper presents a method for the detection of early reflection in the room impulse response using multifractals. The proposed method uses the distribution of Hölder's exponent calculated for the acoustic impulse response for early reflection detection. The obtained results confirm the assumption that impulse response signals exhibit self-similarity, and that fractal theory can be used for the detection of early reflections.