Browsing by Author "Lokki, Tapio, Prof., Aalto University, Department of Signal Processing and Acoustics,Finland"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Parametric Analysis and Reproduction of Indoor Acoustic Environments
    (Aalto University, 2021) Puomio, Otto; Tietotekniikan laitos; Department of Computer Science; Perustieteiden korkeakoulu; School of Science; Lokki, Tapio, Prof., Aalto University, Department of Signal Processing and Acoustics,Finland
    Authentic reproduction of acoustic environments is still one of the major goals in virtual acoustics. One approach to this is to measure the spatial room impulse response of the space and encode it with parametric methods. The spatial sound can be further decoded to different physical output setups, effectively reproducing the space at the point of measurement. This thesis presents five studies revolving around this type of reproduction. The studies address three different subtopics in the field: optimized virtual loudspeaker placement, acoustic scene reconstruction, and plausible spatial sound reproduction. Parametric methods often model static loudspeaker setups that are not aware of the reproduced spatial content. In the worst case, the approach deteriorates the quality of the reproduced spatial sound. The first study presents a method for resolving an optimized loudspeaker setup for a given spatial room impulse response. The listening tests indicated that the optimized setups enhanced the reproduction quality of small rooms, while the differences were harder to distinguish in larger spaces. The rise of virtual reality applications has created a need for a moving listener. Normally, parametric methods are limited to a single position within the acoustic scene, meaning that the listener can only rotate their head in a fixed position. To be able to move from that position, one must reconstruct the measured environment to update the arrival times of early reflections accordingly. Three of the studies present tools that address this topic. Two of them introduce a novel method to reconstruct the scene from multiple parametric spatial room impulse response measurements. The presented method can analyze concave spaces, which has not been possible with earlier methods. The third publication in turn replaces wall material filters with extracted early reflections. The listening tests indicated that the early reflection filters perform the best when they are over 2~ms long. However, optimal filter length could not be determined due to complex interactions with the signal, room, and late reverberation. Virtual reality applications do not necessarily need to reproduce the sound field to the smallest detail. The required level of detail only needs to convince the listener to be within the virtual scene. For this reason, it may be excessive to accurately simulate the space if a sufficient level of detail is achievable by simpler means. The fifth study approaches this idea by introducing a coloration filter to modify a simple room simulation. This aims at coloring the simulated spatial impulse response to resemble the room impulse response utilized to design the filter. The listening test implied that the filter did transform the processed response away from the original one, although the efficiency of the transform was affected by the input stimuli as well as the coloring room impulse response.