Daniel de la Prida Caballero

ABSTRACT

The relevance of the listening test protocol on the detection of Subtle and Just-Noticeable Differences by participants of perceptual experiments in Architectural Acoustics

Listening tests are experimental procedures aimed at assessing how the human auditory system interprets certain auditory stimuli, either individually or by comparison. In this way, listening tests use the human being as an instrument of perceptual evaluation.

In architectural acoustics, listening tests are widely used for a wide variety of purposes. For example, in building acoustics, they are employed to assess the spectral relevance of the sound insulation of construction elements on the perception of annoyance and loudness, among other uses. In room acoustics, these tests have been widely conducted to assess the correlation between standardized room acoustical parameters and the perception of certain room attributes, as well as for the determination of the Just-Noticeable Differences of these classical parameters, among many other aims.

Considering that listening tests use human beings as instruments of perceptual measurement, the methodologies, and more importantly the protocol (i.e., the way in which the stimuli are presented to the participants and how their feedback is collected during the test), adopted for their performance should be precise, robust and homogeneous. However, different studies usually employ very heterogeneous protocols, with different operational powers and robustness. This has led, for example in the determination of the JNDs of classical room acoustics parameters, to different studies with the same goal arriving at drastically different results.

In addition, the data collected from these tests, carried out using very different methodologies, have required the use of wide variety of analysis, which has hindered the comparison and combined use of the results of several studies.

In other areas of sensory perception, such as food theory, there is a large scientific dialogue concerning the operational power and the robustness to bias of different sensory testing protocols. Surprisingly, this is not the case in architectural acoustics, where there are very few publications purely focused on the study of the qualities of testing protocols.

In this talk, the results of a listening test, carried out on 134 participants, are presented. This test was conducted with the aim of comparing the operational power of some overall difference testing protocols, commonly used in room acoustics, such as the Same-Different (SD), ABX and Triangle tests, with others that have proven to have a much higher operational power in other areas of sensory perception, such as the Duo Trio tests with constant reference, both with reference in the first position (CR-DTF), as well as in the first and the middle position (CR-DTFM) and the A-Not/R test, among others. In addition to the operational power, some results are presented concerning the robustness to experimental bias of each of the protocols.

Finally, some practical examples of the use of the most efficient protocols are presented, to carry out listening tests in Architectural Acoustics, showing the benefits of employing Thurstonian Models for the analysis of the results.