Detecting Arrivals in Room Impulse Responses With Dynamic Time Warping

The detection of early reflections in room impulse responses is of importance to many algorithms including room geometry inference, mixing time determination and speech dereverberation. The detection of early reflections can be hampered by increasing reflection width, as the direct sound undergoes reflection and by overlapping of the reflections, as the reflection density grows. This paper investigates the use of Dynamic Time Warping upon a direct sound pulse to estimate the temporal distribution of arrivals in room impulse responses. Bounded Dynamic Time Warping is performed, after an initial correlation of the direct sound with the remaining signal, to further refine the arrival´s location and duration and to find arrivals which may otherwise not correlate well with the un-warped direct sound due to a change in the reflection´s shape. Dynamic Time Warping upon concatenated versions of the direct sound is also used to help find overlapping reflections which may otherwise go undetected. Warping is performed via a set of warp matrices which can be combined together and can also be inverted via a quickly calculated left pseudo-inverse. The algorithm presented is compared to two current methods of arrival detection and is shown to detect a lesser number of spurious arrivals than either of these algorithms.