Regresar

Virtual Reality: A New Approach To Validate Computer Modeling Auralizations By Using Articulation Indexes

Abstract:

In this chapter, a new approach to validate computer modeling auralizations by using articulation indexes is presented. As it is well known, one of the main challenges in validating any technique to produce virtual reality is to define which parameter to consider, since a good (or bad) auralization is a some what subjective evaluation. In this work, the generation of acoustical virtual reality with the proprietary computer code RAIOS is briefly described. The code simulates the room acoustics, including specular and diffuse reflections, and a set of artificial neural networks provides the room's binaural impulse responses at selected points in the audience region. Then, these responses are convolved with anechoicsignals to generate virtual acoustical realities. Details of the measuring protocolto obtain trustworthy articulation indexes are discussed. They are obtained through articulation tests performed with a dictation of monosyllables, recorded in an anechoic chamber, in the Portuguese language. Articulation indexes are also obtained by measuring the binaural impulse responses in the same positions of the selected points used in the numerical simulations. Then, virtual auralizations are produced by convolving the binaural impulse responses with the anechoic signals and the use of equalized headphones. In the sequel, these auralizations are presented to the subjects in order to perform (virtual) articulation tests. Finally, the computer modeling auralizations are also reproduced to the subjects, using the same headphones, and the obtained articulation index results are compared for validation purposes. The main findings in this chapter are the articulation indexes obtained with the computer-based and the measurement-based auralizations. The results show that the articulation index can be used as a metrics to evaluate acoustical virtual realities. Although the proposed metrics do not cover all auralization features, such as sound coloration, spaciality and sound localization, it corresponds to a first-and successful-attempt to establish a parameter to evaluate computer auralizations quality.