Research areas Acoustic Virtual Reality

(Demo section at the end of the page)

Virtual Acoustics

Virtual Reality (VR) is the representation and simultaneous perception of reality and its physical attributes in an interactive virtual environment which is computer-generated in real-time. VR-Technology is used in numerous applications, for instance computational engineering science, mechanical engineering, medicine and architecture. The focus of these applications lies usually on the visual component of the VR environment and acoustics is often added just as an effect without any plausible reference to the physical aspects of the presented scenery. However, it is important to address at least the hearing as well to enforce the feeling of actual presence in the simulated scene (degree of immersion). Especially in the case of architectural/room acoustic applications such as an immersive virtual walk through rooms, e.g. complex of buildings and concert halls, physical models of room- and building acoustics are demanded in order to detect in real-time distinct acoustic effects, e.g., flutter echoes, curved decays and deficient airborne sound insulation.

HYBRID ROOM ACOUSTIC SIMULATION

The determination of the spatial sound field of virtual environments is a difficult task, especially under real-time constraints. Even rather simple situations require quite complex acoustic models and the more the user is allowed to interact with the scenery, the more this complexity increases, e.g., in the case of movable sound sources, movable objects and coupling of rooms. For simulating sound propagation in rooms, a model is required which describes sound emitting sources, physical characteristics of rooms, transmission effects in the case of coupled rooms, and the receiver. A human receiver evaluates these events on attributes which are characterized for example by loudness, coloration, source localization, spaciousness, and the perceived distance to the source. From a system theoretical point of view the perceived sound signal can be represented by the binaural room impulse response (BRIR) convolved with a dry recorded signal of the emitted sound. Thus, the key for a high quality auralization of virtual environments is the computation of BRIRs in real-time in order to produce a plausible sound field at the receiver’s position. In this context, the term plausible means that a number of approximations have to be made due to limited computing power and real time constraints, respectively. However, the resulting sound is not intended to be physically absolutely correct, but perceptively plausible. Therefore, knowledge of the human sound perception and the field of psychoacoustics are essential to find the optimal balance between the needed accuracy on the one hand and the available computation power on the other hand.

The room impulse response can be divided into three parts: the direct sound, the early reflections and the late reverberation. From a perceptive point of view, the direct sound and early reflections affect the localization of sound (precedence effect). A human receiver separately processes the individual reflections (level and angle of impact) and gets an impression for the position and distance to the source. At later times the single reflections are more and more overlapping and the human hearing starts to perform an energetic integration over a certain time slot and angle field. Thus, the early part of the impulse response has to be modeled as accurately as possible whereas a lower time resolution is allowed for the late reverberant sound field. The benefit of the image source model is that it exactly captures every possible specular sound path up to a pre-defined reflection order. This deterministic approach is perfectly qualified for simulating especially the early reflections due to the exact time resolution. Unfortunately, the image source model misses a proper representation of a very important aspect of room acoustics, namely surface and obstacle scattering. It was pointed out by Kuttruff that scattering becomes even a dominant effect in the temporal development of the room impulse response, already from reflections of order two or three. This implies that pure image source modeling would be too much of an approximation of the late part of the room response, which has also been attested by intercomparisons of room acoustics simulation programs.

Better results are achieved by combining image sources with stochastic models for the simulation of the room’s reverberant sound field. In contrast to deterministic image sources, stochastic approaches take into account energetic models for the simulation of sound scattering, e.g., Lambert’s cosine law. Especially the late part of the impulse response is important for a plausible representation of the room’s sound characteristics. Short reverberation times are usually associated to smaller rooms, or vice versa. In addition, spaciousness or spatial impression is strongly related to the listener envelopment and the apparent source width. Bradley showed that especially the level, direction of arrival, and temporal distribution of late arriving reflections have a strong influence on these two important psychoacoustic parameters. Therefore, temporal and spectral directivity information of late arriving sound are to be included to the binaural synthesis in order to create a sound field with a certain performance of realistic physical behavior.

BUILDING ACOUSTIC SIMULATION

To maintain an immersive simulation in the case of complex sceneries, e.g., a complex of buildings, the online auralization of airborne sound from neighboring rooms is mandatory, as it secures the believability of the virtual scene and, thus, the user’s feeling of immersion. For instance the opening and closing of a door to a neighboring room, where a source emits sound at high volume, would be perceived unnaturally if the source is simply switched on and off, respectively. Instead, at least the dominant sound characteristics of airborne sound, i.e. the level (low) and coloration (dull), have to be auralized. In the case of coupled rooms, the process of sound transmission between two rooms includes the room acoustic simulation of the sound field in the source room, the modeling of the excitation and radiation of bending waves in the walls, and a room acoustic simulation of the sound field in the receiver room. Fortunately, the airborne sound transmitted from an adjacent room into the receiver room through direct and flanking room components can well be auralized by using secondary point sources located in the centre of each transmitting structural element of the receiver room. The benefit of this kind of building acoustics auralization method is that it requires only the sound reduction index, R, of each construction element as additional input data, where the single indices can be measured after ISO 140, while their overall performance in the building structure, including direct and flanking transmission, can be computed according to standardized simulation models, e.g., the European Standard EN 12354. The auralization concept of sound transmission is shown in the figure below.

Simple room-to-room situation, where source and receiver are located in two different, but coupled rooms. The source room is acoustically substituted for 8 secondary point sources located in the receiver’s room, i.e., direct and first order junctions are taken into account.

RAVEN

The hybrid room acoustics simulation software RAVEN (Room Acoustics for Virtual ENvironments) is currently being developed at the Institute of Technical Acoustics, RWTH Aachen University, Germany, which takes into account all criteria mentioned above. RAVEN combines deterministic image source method with a stochastic ray tracing algorithm in order to compute impulse responses in real-time which reach state-of-the-art room acoustics simulation standards. RAVEN is incorporated into the 3D sound rendering system of RWTH Aachen University’s CAVE and is used for any room acoustics simulation as a networked service.

An (out-dated) overview of the 3D sound rendering system is given in:
Virtual Reality System with Integrated Sound Field Simulation and Reproduction

Video Downloads

Video: Auralization of coupled rooms This video demonstrates the interactive auralization of airborne sound insulation between two rooms.	Video: Gesture-based design of a concert hall This video demonstrates an interactive gesture-based design of a concert hall.
	#
Video: Multimodal simulation of natural sound sources This video demonstrates the multimodal simulation of a sax player (motion/directivity). The player's motion was captured by trackers installed in our CAVE. The motion data is not post processed. Occuring artefacts come from tracking drop outs.	Video: 3D-Visualization of simulation results This video demonstrates RAVEN's visualization module. The module features several visualization modes for room acoustical paramters and other energy quantities such as the propagation of energy particles.

Audio Downloads

Dry signal. This audio file was recorded in an anechoic chamber.	Auralization: Classroom Binaural signals, use headphones!	Auralization: Concert Hall Binaural signals, use headphones!

Pictures

Cave Impressions	Cave Impressions	Cave Impressions	Cave Impressions
3D-Visualization: Center time	2D-Visualization: Center time	3D-Visualization: Definition	2D-Visualization: Definition
3D-Visualization: Particle Propagation (Ray Tracing)	3D-Visualization: Particle Propagation (Ray Tracing)	3D-Visualization: Particle Propagation (Ray Tracing)	3D-Visualization: Particle Propagation (Ray Tracing)
3D-Visualization: Specular Reflections (Image Sources)	3D-Visualization: Scene Selector	3D-Visualization: Directivity Trumpet (500 Hz)	3D-Visualization: Directivity Trumpet (2 kHz)

Cooperation Partner:	ITA is member of: