Peter Svensson

Analysis of SSIM performance for Digital Cinema applications

Visual quality is one of the most important issues in Digital Cinema applications, and the most practical method to measure visual quality is using objective metrics. Several objective metrics that take into account perceived visual quality have been developed in the past few years; one of them is based on the structural similarity paradigm. To analyze the performance of these objective metrics for digital cinema applications, a subjective visual quality assessment in digital cinema environments using contents from the Digital Cinema Initiative (DCI) Standard Evaluation Material has been conducted in a previous study. Since in practice digital cinema utilizes only high quality imagery, in this paper we analyze further a subset of highquality stimuli from the previous study.

Room acoustic prediction and auralization based on an extended image source model

During the 1980s, an exciting new technique was developed that put great demands on prediction methods. This new technique, called ‘‘auralization’’ in an analogy to visualization, enables the designer of a concert hall to get an aural impression of the hall before it has been built. By the calculation of highly detailed octave‐band echograms, the use of signal‐processing techniques and measured head‐related transfer functions, a binaural room impulse response can be synthesized. This response is convolved with anechoically recorded music to give an impression of how the music would sound in the completed hall. This paper describes auralization using the image source model. This model, as well as others, has its share of short comings. In an attempt to compensate for some of these, the basic model has been extended. The two most important extensions are the treatment of scattering and the use of ray tracing to estimate the late part of the echogram. The combined method enables reasonably fast calculations,...

Influence of electroacoustic parameters on the performance of reverberation enhancement systems

The computer simulation of electroacoustic reverberation enhancement systems for variable acoustics is presented. All involved room impulse responses are predicted using the image source model, complemented by a diffuse field extension for the late part. The electronic units included in the reverberation enhancement system are also specified by their impulse responses. The corresponding frequency responses are processed using a matrix formulation which yields the total room impulse response of the hall with the reverberation enhancement system installed. From this, measures which describe the system performance can be calculated. In the simulation study presented, a typical medium‐sized hall which seats 450 persons has been modeled and nine different two‐channel configurations have been simulated. The current simulation approach proves more flexible than previous methods since also the early part of the total impulse response can be predicted. Transducer directivities and electronic reverberation can be included and it is shown that these, along with the system delay, to a large extent influence the early part of the total impulse response. The risk of feedback is also discussed.

Subjective and objective assessment of some reverberation enhancement systems

How much of a difference does active or passive variation of a rooms acoustical conditions make? Electronic architecture and reverberation enhancement have become common in the pursuit of better acoustics in multipurpose auditoria by active means. New systems show promising results. The properties of several different installations both in the U.S. and Europe have been investigated. This has been done both objectively by measurement of omnidirectional and binaural impulse responses and subjectively by the use of binaural recording. The binaural recordings used anechoic music replayed in stereo on stage. These recordings were used in pair comparison tests to investigate the dimensions of audible difference between the halls, with and without active or passive treatment of the room response. The pair comparison tests were evaluated using multidimensional scaling. The results show that some systems are able to modify the acoustical conditions to a very large extent. Differences between various system settin...

Frequency‐domain diffraction for edges of arbitrary length with efficient numerical integration

Frequency‐domain edge diffraction from an infinite wedge insonified by a point source has been widely studied, with various analytical solutions dating back nearly a century. In this talk we present an alternative frequency‐domain solution which can be used for finite as well as infinite edges. The expression, given as a line integral along the diffracting edge, is derived from an analytical time‐domain expression presented in [U. P. Svensson et al., J. Acoust. Soc. Am. 106, 2331‐2344 (1999)]. The new formulation is shown to be equivalent to an exact, analytical, contour‐integral formulation for the infinite wedge via a variable transformation. Results for various finite and infinite edge cases will be presented, and numerical integration using an efficient quadrature method for highly oscillatory integrals will also be discussed.

The generation of binaural signals from a regularly sampled soundfield using a beamforming approach

Auditorium designs can be evaluated prior to construction by numerical modeling of the design. Subjective assessment of the design requires auralization of the computed soundfield at a desired listener position and the binaural format is dominant for representation of the soundfield. For high accuracy the modeling is based on solving the wave equation numerically, which typically produces the sound pressure on a rectangular grid. Such a sampled soundfield is not straightforward to convert to a binaural format. This paper investigates binaural conversion from the sound pressure at a selected number of grid points using a least squares beamforming approach. Low‐frequency axisymmetric emulations are derived assuming a solid sphere model of the head, and an open spherical array of 640 microphones is used to emulate ten measured HRTF data sets from the CIPIC database for half the audio bandwidth. The microphone positions might be displaced from the exact spherical surface to fit the mesh sample points. The sph...

The effect of a preprocessing approach improving speech intelligibility in reverberation considering a public‐address system and room acoustics

This study evaluates a preprocessing approach for reducing reverberation effects when the input signal is not ideal, dry speech, but rather a realistic speech signal picked up by a close microphone in a room. And this study shows how the situation affects the input and the further approach compared to a dry signal as the input. Steady‐state suppression, as described by Arai et al. [Acoust. Sci. Technol. 23, 229–232 (2002)], was used as a preprocessing approach that processes a speech signal before it is radiated from loudspeakers. A lecture was simulated in two different halls (reverberation times of 1.2 and 1.8 s) in which public address systems were installed. The simulation software CATT‐Acoustic was used and impulse responses were calculated for the input to the preprocessing approach and for a listener position. Stimuli for a syllable identification test were prepared by convolving speech signals with the calculated impulse responses. Speech signals were given with and without steady‐state suppressio...

Computation of impulse response for stage‐house geometries with use of edge‐diffraction models

A mathematical model following diffraction studies by Medwin [J. Acoust. Soc. Am. 72, 1005–1013 (1982)] is developed to calculate the impulse response of a three‐dimensional ‘‘stage house’’ geometry in a baffle. The model includes multiple diffraction components in addition to specular reflection, and it forms a more complete approach than current room impulse‐response predictors based on geometrical acoustics. Thus it may also be suitable for more accurate auralization, particularly at low frequencies. A comparison is made with results from a ‘‘cone‐tracing’’ computer model [B. Dalenback, J. Acoust. Soc. Am. 100, 899–909 (1996)] and results from physical scale‐model measurements. [Work supported by Axel & Margaret Axson Johnson’s Foundation, Sweden.]

Room reverberation measurement by conventional methods and by the MLSSA measurement system

Experiments have been made to compare the measured T60 values measured by conventional analysis equipment to those measured by equipment using the maximum length sequence (MLS) correlation technique such as the commercial MLSSA system. Experiments have also been done using MLSSA impulse response results in conjunction with signal postprocessing in MatLab. Results show that the MLSSA system yielded values which differed by up to 10% from those obtained by conventional techniques. The reasons for this are not entirely clear. MatLab postprocessing also allows signal postprocessing to improve the dynamic range of measurement and the implementation of routines for more specialized room acoustic measures. [Work supported by the Swedish National Council for Building Research.]

Computer simulation of electroacoustic reverberation enhancement systems.

Computer simulation of electroacoustic reverberation enhancement systems (RES) for auditoria is described and compared with actual measurements. The simulation proceeds in two stages: First, the room impulse responses (from the source and all loudspeakers to the receiver and all microphones) are predicted using the mirror image method, which is then complemented by a diffuse field extension for the late part. The electronic units (amplifiers, delays, and reverberation units) are specified by their impulse responses. Second, the impulse responses are convolved and added using a matrix formulation. This yields the total room impulse response of the hall with the RES installed and objective criteria can then be calculated. Auralization is also possible by convolving the total impulse response with anechoic source signals. More complex systems can be studied than was possible with previous methods since transducer directivities and individual channel reverberation can be included. It is shown that these param...