Martin Pollow

Acoustic centering of sources measured by surrounding spherical microphone arrays

The radiation patterns of acoustic sources have great significance in a wide range of applications, such as measuring the directivity of loudspeakers and investigating the radiation of musical instruments for auralization. Recently, surrounding spherical microphone arrays have been studied for sound field analysis, facilitating measurement of the pressure around a sphere and the computation of the spherical harmonics spectrum of the sound source. However, the sound radiation pattern may be affected by the location of the source inside the microphone array, which is an undesirable property when aiming to characterize source radiation in a unique manner. This paper presents a theoretical analysis of the spherical harmonics spectrum of spatially translated sources and defines four measures for the misalignment of the acoustic center of a radiating source. Optimization is used to promote optimal alignment based on the proposed measures and the errors caused by numerical and array-order limitations are investigated. This methodology is examined using both simulated and experimental data in order to investigate the performance and limitations of the different alignment methods.

Variable Directivity for Platonic Sound Sources Based on Spherical Harmonics Optimization

Sound sources for measurements in room acoustics are of omni-directional type, in general. With respect to auralization applications, an omni-directionally measured room impulse response may not be the ideal choice since it does not represent the real life situation playing an instrument in the room. To achieve the directivity of a real source (like a musical instrument or human voice) with a technical sound source (like a loudspeaker) requires either to copy the entire body and the sound radiation (i.e. the surface velocity) of that particular source or to reproduce the directional pattern of the radiation using a multiple source configuration like a dodecahedron or icosahedron loudspeaker array with independent excitation of each transducer. The advantage of the latter method is obvious since one single source is able to provide a large variety of different directivities by simply changing the excitation profile. To maintain the appropriate excitation of each individual transducer, different approaches can be made. In this paper a method is described using a discretely measured target radiation pattern to calculate the frequency dependent excitation signals for each transducer. Hereby, the directivity pattern of the source transducers and a phase optimization of the energy averaged radiation of musical instruments are used. The advantage of this method is a very flexible computation that is able to match the radiation pattern at the points of measurement very well. The resulting input filters for the platonic sound source can be used either for a real time convolution or offline processing of measured signals.

Including directivity patterns in room acoustical measurements

Room acoustical measurements according to ISO 3382 require source and receiver to be of omnidirectional sensitivity. Therefore, radiation patterns of natural sources and receivers (although audible) are not accounted for when using the obtained room impulse responses (RIRs) for room acoustic analysis or even auralization. In order to include this spatial information in the RIR, it is necessary to measure the RIR for each pair of radiation patterns of source and receiver. This could be done by electronic beamforming during the measurement using array systems or by mechanical modification of the transducer (as, e.g., a dummy-head with its corpus). In this contribution, an alternative approach is shown, using the superposition of a set of sequential measurements done with a spherical sound source. At the cost of longer measurement times, the obtained data can be used universally to synthesize RIRs of arbitrary directivity up to a certain maximal spatial resolution, as long as the room is considered as a line...

Synthesis of room impulse responses for arbitrary source directivities using spherical harmonic decomposition

The acoustics of rooms can be accurately described by the room impulse response, stating the temporal succession of the room reflections encountered for a specific transfer path in the room. The measurement is usually performed with a single channel spherical loudspeaker—ideally with an omnidirectional radiation pattern—providing an objective measure for a given combination of a source and receiver position in a room. The measurement result can be used to auralize sources in the measured room by convolution with a dry signal or to derive standardized room acoustical parameters. However, every acoustic source has a distinct frequency dependent radiation pattern, which is to some extent responsible for the perceived characteristics of the sound source. The conventional measurement of room impulse responses does not account for this effect. Hence, a method is proposed to obtain room impulse responses for specific radiation patterns by superposing measurement results of an electro-acoustic sound source with a known radiation pattern for different orientations. The limitations of the method are examined in terms of frequency range and spherical harmonic order. The assumption of an LTI system is studied by special measurements. Finally results of measured room impulse responses for a target source are compared to the synthesis results obtained by the proposed method.

Spherical harmonics based HRTF datasets: Implementation and evaluation for real-time auralization

HRTF filters used in most binaural synthesis application stem from a discrete set of either measured or simulated far-field data. While this data format allows fast filter generation times and is fairly straightforward to use, correct filter representation is only possible for the measured points while an interpolation is needed between the points. As these interpolated filters are not a physically correct representation of the HRTF, this approach is not suited for the auralization of very small head movements that humans tend to do to improve localization [1]. For the auralization of sources in the vicinity of the listener the near-field HRTF is important [2]. With discrete HRTF data sets, multiple measurements for different distances have to be combined to account for this.

Representation of the musical instruments directivity using dodecahedron loudspeakers.

Room‐acoustical measurements in general are performed with omnidirectional sound sources. With respect to auralization, such an impulse response may not be ideal since it does not represent the situation playing an instrument in the room. To achieve the directivity of a real source (such as an instrument or human voice) with a technical sound source (a loudspeaker) requires either to copy the body and the surface velocity distribution of that particular source or to reproduce the directional pattern of the radiation using a multiple source configuration like a dodecahedron loudspeaker array with independent excitation of each transducer. The advantage of the latter method is obvious since one single source is able to provide different directivities by changing the excitation profile. To maintain the appropriate excitation of each individual transducer, different approaches can be made. The method described here uses spherical harmonics decomposition of the target radiation pattern and a subsequent calcula...

Variable Source‐Directivity Using Dodecahedron‐Loudspeakers

For room‐acoustical measurements dodecahedron loudspeakers are commonplace to achieve a uniform directivity. Therefore all transducers are fed with the same signal. If the signals for the twelve transducers are individually adjustable, the variation of amplitude and phase offers the possibility to achieve a predefined directivity. The goal is to calculate the twelve frequency dependent amplitude‐ and phase‐coefficients for any given directivity with the least possible error. A simple approach like superposition unfortunately does not reveal a correct result, since all transducer interact with each other. The decomposition of spherical functions into spherical harmonics, however, leads to an analytic solution for the prediction of the sound radiation. The acoustical components ‐ like sound pressure and sound velocity ‐ are split up into weighted, orthogonal base functions which can be combined in a way that the mutual coupling between different membrane vibrations is respected. Under these conditions compl...

A Database of Anechoic Microphone Array Measurements of Musical Instruments

Violin (classical) Violin_historical 430 Hz G3-C7 G3-C7 Normal of the soundboard aiming at Mic#25, Scroll aiming at between Mic#02 and Mic#03 Daniel Deuter 200905-18 14:15 15:15 Instrument: Neil Kristóv Értz (1998, rev. 2003), Copy of Guaneri; Bow: Violin Bow (50 g), René Groppe (Metz, 1997), Copy of baroque model (17th century); Strings: E String: Catgut (Damian Dlugolecki, 0.62 mm); A String: Catgut (Nicholas Baldock, Kathedrale Strings, 0.74 mm); D String: Catgut (+0.25 mm Silver, 1.12mm, Nicholas Baldock, LuxLine); G String: Catgut, Silver, diameter unknown

Interfacing spherical harmonics and room simulation algorithms

Room acoustic simulation by using geometrical acoustics is usually implemented with binaural receivers. Wave models such as FEM are easily applicable with binaural interfaces as well. This way, however, the signals are restricted to a specific set of HRTF, and a tedious task is to adapt the results to a proper reproduction system with very limited possibilities of listener individualization. With a more general interface such as spherical harmonics, room acoustic spatial data could be created in intermediate solutions. In post-processing this can lead to various binaural representations or to reproduction with Ambisonics (Dalenback, ICA 1995). In this paper it is discussed how standard routines in geometrical acoustics must be changed in order to implement multi-channel spherical microphone arrays. Furthermore, the corresponding output data can be multi-channel time signals or temporal SH coefficients or any other suitable spectral format. The amount of data and signal processing affects CPU time and memo...

Reciprocal binaural room impulse response measurements

Multi-channel spherical loudspeakers have been introduced in shapes of cubes, dodecahedra, or higher-order discrete representations of spheres. In this contribution a spherical source with a partial Gaussian distribution of 28 channels is presented. With sequential measurements and rotation of the sphere a radiation of effectively 23rd order of spherical harmonics can be obtained. Accordingly directional patterns of not only sound sources but also of receivers such as HRTF can be modeled in detail up to quite high frequencies. The high order of spherical harmonics allows investigation of individual differences of pinna cues. When applied in a reciprocal measurement of room impulse responses in performance venues, an almost perfect omnidirectional microphone on the stage and an HRTF source in the audience can be used to study spatial room acoustic parameters such as early lateral energy fractions, late lateral strength and IACC of dummy heads and individuals. This is obtained by post processing of just one set of multi-channel impulse responses in the venue. Opportunities and challenges of this approach will be discussed. Multi-channel spherical loudspeakers have been introduced in shapes of cubes, dodecahedra, or higher-order discrete representations of spheres. In this contribution a spherical source with a partial Gaussian distribution of 28 channels is presented. With sequential measurements and rotation of the sphere a radiation of effectively 23rd order of spherical harmonics can be obtained. Accordingly directional patterns of not only sound sources but also of receivers such as HRTF can be modeled in detail up to quite high frequencies. The high order of spherical harmonics allows investigation of individual differences of pinna cues. When applied in a reciprocal measurement of room impulse responses in performance venues, an almost perfect omnidirectional microphone on the stage and an HRTF source in the audience can be used to study spatial room acoustic parameters such as early lateral energy fractions, late lateral strength and IACC of dummy heads and individuals. This is obtained by post processing of just one...