Diemer de Vries

Wave field synthesis and analysis using array technology

The concept of wave field synthesis (WFS) was introduced by Betkhout in 1988. It enables the generation of sound fields with natural temporal and spatial properties within a volume or area bounded by arrays of loudspeakers. Applications are found in real time performances as well as in reproduction of multitrack recordings. A logic next step was the formulation of a new wave field analysis (WFA) concept by Berkhout in 1997, where sound fields in enclosures are recorded with arrays of microphones and analyzed with postprocessing techniques commonly used in acoustical imaging. This way, both the temporal and spatial properties of the sound field can be investigated and understood. WFS and WFA meet in auralization applications: sound fields measured (or modeled) along arrays of microphone positions can be generated by arrays of loudspeakers for perceptual evaluation.

Spatial fluctuations in measures for spaciousness

In room acoustics, several measures have been defined that are supposed to quantify the apparent source width (ASW) in a hall, being one of the perceptual cues related to spaciousness. The most common ones are the lateral energy fraction (LF), i.e., the ratio between lateral and omnidirectional early energy, and the interaural cross correlation coefficient (IACC), all to be calculated from measured or simulated impulse responses. [Several versions of the LF are known in literature, having different names, generalized here as lateral energy fraction.] According to a method proposed by Berkhout et al. [J. Acoust. Soc. Am. 102, 2757–2770 (1997)], for a fixed source position impulse responses have been measured along an array of closely spaced microphone positions in several halls. The above measures, when calculated from these impulse responses, show large fluctuations with small variations in microphone position due to interference of the different components of the wave field to which the human ear is appa...

Extraction of 3D information from circular array measurements for auralization with wave field synthesis

The state of the art of wave field synthesis (WFS) systems is that they can reproduce sound sources and secondary (mirror image) sources with natural spaciousness in a horizontal plane, and thus perform satisfactory 2D auralization of an enclosed space, based on multitrace impulse response data measured or simulated along a 2D microphone array. However, waves propagating with a nonzero elevation angle are also reproduced in the horizontal plane, which is neither physically nor perceptually correct. In most listening environments to be auralized, the floor is highly absorptive since it is covered with upholstered seats, occupied during performances by a well-dressed audience. A first-order ceiling reflection, reaching the floor directly or via a wall, will be severely damped and will not play a significant role in the room response anymore. This means that a spatially correct WFS reproduction of first-order ceiling reflections, by means of a loudspeaker array at the ceiling of the auralization reproduction room, is necessary and probably sufficient to create the desired 3D spatial perception. To determine the driving signals for the loudspeakers in the ceiling array, it is necessary to identify the relevant ceiling reflection(s) in the multichannel impulse response data and separate those events from the data set. Two methods are examined to identify, separate, and reproduce the relevant reflections: application of the Radon transform, and decomposition of the data into cylindrical harmonics. Application to synthesized and measured data shows that both methods in principle are able to identify, separate, and reproduce the relevant events.

Spatial sound reproduction with wave field synthesis

Wave field synthesis is a reproduction technique developed at TU Delft, that enables the generation of high‐quality three‐dimensional spatial sound fields. The benefit of the method is that spatial impressions are highly independent of the position of the listeners within a large listening area. In short, the method uses a limited number of audio channels that are reproduced by generating plane and spherical wave fields with arrays of loudspeakers that surround the listening place. Applications include spatial sound reproduction in the home and in cinemas, sound reinforcement in theaters, teleconferencing with large video screens, and variable acoustics.

Wave Field Synthesis: History, State-of-the-Art and Future (Invited Paper)

In this paper, wave field synthesis (WFS) is discussed in a rather tutorial way. The principles of WFS - a concept to realize natural spatial acoustics in a `sweet spot?- less listeners area - are explained. Recording techniques for WFS reproduction are discussed. It is concluded that, after ample developments in the framework of the CARROUSO project, the Delft concept has by now been accepted worldwide for use in various applications. Finally, the state-of-the-art and expectations for the future are discussed.

Modelling and order of acoustic transfer functions due to reflections from augmented objects

It is commonly accepted that the sound reflections from real physical objects are much more complicated than what usually is and can be modelled by room acoustics modelling software. The main reason for this limitation is the level of detail inherent in the physical object in terms of its geometrical and acoustic properties. In the present paper, the complexity of the sound reflections from a corridor wall is investigated by modelling the corresponding acoustic transfer functions at several receiver positions in front of the wall. The complexity for different wall configurations has been examined and the changes have been achieved by altering its acoustic image. The results show that for a homogenous flat wall, the complexity is significant and for a wall including various smaller objects, the complexity is highly dependent on the position of the receiver with respect to the objects.

Fluctuation of room acoustical parameters on small spatial intervals

The Delft Acoustics group has introduced array technology in room acoustic measurement practice: instead of at a limited number of ‘‘representative’’ places, impulse responses are measured along an array of closely sampled microphone positions. By displaying the responses as a visual entity, the wave character of the sound field is clearly revealed. The data form a base for further processing, enabling the separation and analyzing of the different wave‐field components. The sound fields in several concert halls have been measured this way. From impulse responses recorded with 0.05‐m spatial separation, the common room acoustical parameters have been determined. It appears that, even between the ten positions in front of one and the same seat, unexpected fluctuations occur, e.g., when measured in octave bands, the clarity index varies over 1 dB on this interval, the lateral energy fraction 0.2 (i.e., between 0.1 and 0.3). For the broadband versions, these variations are approximately halved, still being significant. These results explain why the predictive value of these parameters for the perceptual quality of the acoustics on a (group of) seat(s) is limited. Instead of at one point, the values at several adjacent points should be considered.

Usage of the Fourier transform as an invertible extension to the Radon transform with application to wave front extraction

The Radon transform is used in many applications in order to detect predefined geometrical structures, for instance curves in a given two-dimensional dataset. It is a mapping between a data space and a model space; the coordinates of a point in the latter correspond to the parameters of a geometrical structure in the former. The construction of an inverse operator is usually not straightforward, it is however possible to obtain numerical approximations. In this paper, an extension to the Radon transform is proposed which simplifies inversion by using a Fourier transform along the input curve instead of a line integral. Thereby, the model space is extended by one dimension containing the Fourier coefficients, from which information about the amplitude variation along the input curve can be extracted. The method is successfully applied in order to extract wave fronts from acoustical impulse response measurements taken on a circular microphone array.

Modelling of the cochlea response as a versatile tool for acoustic signal processing

The inner ear or cochlea processes the acoustic signals that enter the oval window into a specific time‐frequency pattern. Many acoustic signal processing methods are based on this behaviour. A fundamental method is to calculate this time‐frequency response by solving the differential equation of the movement of the basilar membrane, followed by a visualisation of the excitation patterns in a time‐frequency plot. For that purpose Continuity Preserving Signal Processing (CPSP) is a promising method. In the presentation an overview will be given of a project that is carried out by TUD (University of Technology Delft) together with RUG (University of Groningen) being sponsored by STW (Dutch Technology Foundation). The project divides into four sub‐projects which are closely related: Automatic Keyword Spotting, Machine Analysis and Diagnostics, Speech Intelligibility Enhancement for Hearing Aids and Quality Assessment of Room Acoustics. Results that have been obtained in the project will be summarised. Detailed results of the sub‐projects will be presented in separate presentations.

Determining acoustical parameters using cochlear modeling and auditory masking

The acoustical qualities of a concert hall or any other room are generally expressed using acoustical parameters determined from impulse responses. From microphone array measurements it turned out that these parameters can fluctuate severely over small distances, whereas the perceptual cues for which these parameters are supposed to be a measure remain constant. This means that a local parameter value has a very low predictive value for acoustic quality. In this research, cochlear modeling techniques and simulations of auditory masking effects have been applied to model human hearing. These techniques together model various stages in the auditory path, like the movement of the basilar membrane inside the cochlea and mechanisms inside the brains. It turns out that determining acoustical parameters using this representation leads to results which show much less spatial fluctuations, and are closer to human perception.