Contribution
Comparison of Noise Detection Methods for Exponential Sine Sweep Measurements
* Presenting author
Abstract:
Spatial room impulse responses (SRIRs) provide a variety of information about the acoustic properties of a room. Datasets of measured SRIRS may be used e.g. for room acoustic analysis, the validation of room acoustic simulations, as training and test datasets for data-driven processing, or directly for auralization. These datasets can reach a high degree of complexity in terms of the number of individual data and high spatial resolution. As these kinds of measurements are time-consuming, automated measurement procedures are occasionally used. In spite of the advantages of these methods, they also pose new challenges for error detection, as the datasets they generate are too large to be checked manually.This paper compares two methods for noise detection using large data sets collected under field conditions. One method is the impulse noise detection according to Guski and the other is the correlation-based Rule of Two according to Prawda et al.The results are evaluated and the advantages and disadvantages are worked out with particular regard to automated measurements. The possible sources of error are discussed, some of which can be attributed to the inherent noise of the robot platform used.