Journal of Sonic Studies, volume 6, nr. 1 (January 2014)Iain McGregor; Phil Turner; David Benyon: USING PARTICIPATORY VISUALISATION OF SOUNDSCAPES TO COMPARE DESIGNERS’ AND LISTENERS’ EXPERIENCES OF SOUND DESIGNS
5. Further work
Although the ten sound designers who took part in the final study agreed that visualisation of the soundscape allowed them to compare a sound design with the experiences of listeners, further work has been identified. Research needs to be conducted on the internal validity of each attribute; these attributes will be tested individually in order to establish the correct scale and be provided with easily understood descriptors. The attributes associated with the physical properties of sound events, such as Spatial, Temporal, Dynamics, and Spectral might be expected to be more consistent across listeners than the more subjective attributes of Interaction, Aesthetics, Emotions, or Content. However, this was not always the case, which can be seen in the significant level of reliability for Content and Emotions in Sonification and Radio Drama.
Of the 7 attributes that are retained, 6 had a single method of visualisation chosen by the majority of the designers. There was only a low level of agreement for the visualisation of the Dynamics attribute, and varying the dimensions of the shape was chosen by only four out of ten of the designers. Alternative methods of displaying this attribute will be researched and trialled with designers.
Individual visualisations of each of the listeners’ responses vary from being almost identical to disparate, with the majority being similar. The designers’ intentions are not always markedly different from the listeners’ experiences, as in Radio Drama and Sonification, where two thirds of the attributes are rated identically. In contrast, there are obvious differences between designers’ intentions and the listeners’ experiences for both Soundscape Composition and Auditory Display, with only a third of the attributes being rated identically. In order to identify if it is the act of design or the expertise of the designer that introduces the differences in responses, each designer could analyse the sound designs that they did not create themselves, and the results could be compared to the listeners’ experiences.
The low level of agreement for Simulation and Auditory Display might suggest that greater levels of agreement may be achieved if non-experts were to design for non-experts, and experts designed for experts. However, it could also be argued that the approach might be used as part of an iterative design process, where experts become more attuned at designing for non-experts and adapt their designs accordingly. The similarity between the designers’ and listeners’ responses for Radio Drama and Sonification illustrate that responses are not always disparate.
Studies will also be conducted to establish the most appropriate length and level of complexity for listeners to experience. There was not an easily identifiable relationship between length and complexity of a sound design with regards to the level of awareness of sound events. A balance between reliance on listeners’ memories and the level of awareness needs to be made.
A method of highlighting only the differences between a sound designer’s intentions and the listeners’ experiences within a single visualisation was suggested by one of the designers. A possible way of showing the differences on a single visualisation might be to overlay the listeners’ responses onto the designer’s, omitting all of the attributes that are identical, whilst still retaining the code in order to indicate the sound event’s presence (see Table 19 and Figure 23). In this example, each sound event has been given a different colour in order to aid identification. The sound events are located on the grid according to the designer’s responses, and the arrows indicate the positions according to the listeners. The designer’s response is on the left hand side of each object, with the listeners’ on the right.
An alternative approach might be to convey the results through sound, either through a sonification or a sound design. A sonification that conveys agreement for each attribute could be running concurrently. If there is agreement, then there is no additional sound. If there is a differing response between designer and listeners, then an appropriate tone or alert could be played, according to the attribute. Another solution could be to alter the sound design according to the listeners’ experiences. Attributes that represent physical properties of sound, such as Spatial, Temporal, Spectral, and Dynamics would be relatively easy to alter so that the design that more closely matches the listeners’ experiences could be made available for the designer to experience. Awareness would be the easiest to convey, as it would only require the removal of a sound event. Attributes such as Emotion and Content would be more difficult to convey, however, through analysis of successful matches, it might prove possible to alter the sound design so that the designer can experience the listeners’ perspective. Although the study contributes in general to the design of sound for any application, there are specialist applications which might benefit as well. These applications include those for the visually impaired, or mainly out of the user's view, without a screen, or in high performance environments in which the misinterpretation of audio cues might have serious repercussions.
The aim of this work is to enable sound designers to evaluate their designs’ intended meaning with their listeners. In this paper we have reported on an evaluation of the soundscape visualisations, involving ten designers and 100 listeners. The study provided examples of how the visualisations could be used by designers to compare their intentions for a sound design with the experiences of listeners. The need to visualise sound designs has not been established in this paper, a method has merely been presented, and has not been compared to other forms of comparison. The ten designers who took part in the study all agreed that soundscape visualisation is an effective way of evaluating soundscapes for their sound designs. They proposed a number of modifications, the most important being the reduction in the number of attributes needed to describe the salient features of a sound event. Before any assessment about the suitability of visualisation for sound design in general is made, it is important to develop the method further and test it in a more systematic manner, as there is little evidence that the approach could describe a general range of sound design practices. Revisiting the work of authors such as Barrass (Barrass 1996), should help provide descriptors more robustly suited for the design of sonifications and auditory displays in general.
Research continues to refine the visualisations, based on the 7 attributes favoured by the designers. The reduction in the number of attributes should reduce the cognitive demand on those interpreting the maps, as the number of symbols has been reduced from 13 to 7. We hope that with a smaller set of attributes and clearer notation for the visualisations we can produce a method for the evaluation of soundscapes that could become a standard part of interaction design. At present, a sound design expert needs to draw the visualisations, but with the more focused set of attributes, tool support, and a database of sound design patterns, we are confident that the benefits of soundscape visualisation would be easily available to designers. We are of the opinion that using a visual form to represent soundscapes could make an interaction designer’s job easier and more efficient.