Room Surface Estimation Using Reflection Coefficients Measured In-Situ

Abstract

Room surface estimation is the process of estimating and characterizing the surfaces of a room from the measurement provided by an array of microphones. In any normal room, wavefront radiated by the source reaches the microphones after reflecting from the surfaces of the room. On performing the post processing on this recorded signal, depending on the type of source signal used; we obtain the unique signature of the room called the room impulse response (RIR). Historically room impulse responses have been used to calculate the acoustical parameters. These are generally used for objective evaluation of the rooms. Here is an effort to extract more information out of the RIR by understanding the physics of it.

In this thesis the reflection from the surfaces is employed as useful information and methods to use this information are presented. The reflective information can be used in the determination of the reflection coefficient of the surface. The reflection coefficients of four common room materials are obtained from their respective impulse response at oblique incidence. The obtained reflection coefficients are studied for the classification of materials.

The problem of estimating the room surface using the reflection coefficients has been approached in a systematic way, accounting for parasitic reflections and background noise. The room impulse response is obtained using the sine sweep signal transmitted through the speaker attached to one microphone array, and recorded by another microphone array. The recorded signal was used to identify possible surfaces using clustering on the obtained reflection coefficients. The classification so obtained is compared with the ground truth to calculate the performance and practicability. /Kir11