Search Results

Automotive audio systems require signal processing to address losses of spatial rendition of stereo and multichannel sources due to off-centerline listening and the proximity of cabin walls where the speakers are located. The digital signal processing is optimized for each vehicle type and audio option by skilled tuning engineers. They spend the majority of their tuning time on spatial issues when preparing a vehicle. One reason for this is that, until now, there has been no measurement of imaging performance. Engineers have relied on the indirect measures of arrival delay, spectrum and their own hearing to balance the several speakers likely to be contributing to perceived direction, distance and ambience. This paper describes a new module for a measurement system widely used in automotive audio. In addition to assisting the tuning engineer, the system documents audio performance in Tonal Balance, Maximum Output, Distortion, and now, Spatial Rendition.

It has been know for many years that the highest level of audio performance is achieved at the hands of an expert tuning engineer. It is recently that automotive signal processing options have become overwhelming in their number and variety. In addition, the number of vehicles using model-specific equalization has grown with the introduction of digital signal processing (DSP). It has been observed that much engineering time is consumed before arriving at the perfect tuning. To make the process a bit more “cookbook,” a structured approach is presented.

A loudspeaker's low frequency parameters can be accurately estimated by measuring individual components without the need to assemble them into a working unit. Using this development approach, much time can be saved by optimizing one component at a time rather than building an entire speaker for each iteration. To do this, one must be able to measure the relevant physical characteristics of each component (the electro-mechanical parameters). It is also necessary to be able to predict electro-acoustic performance from the electro-mechanical parameters.

Measurement of acoustic frequency response is extended to include aspects of speech and music perception in rooms. An apparatus is described that combines established and new techniques in a computer based measuring system. The resulting objective plot has a better correlation with subjective assessment than previous methods. The flexibility of the measurement platform encourages future development.

Unacceptable distortion, sounding like a “blat” on bass notes was reported for some models of 4 X 6-inch loudspeakers in certain 1992 prototype vehicles. A study was initiated to find the cause of this distortion and to develop a specification to limit it to acceptable values. The cause of the distortion was determined to be a rapid increase in suspension stiffness at large cone excursion. The specification is based on the sum of the weighted acoustic amplitudes of the 4th through 10th harmonics of a sine wave input at 8.0 volts.

A simple method for measuring the peak linear excursion capability (Xmax) of loudspeaker drive units is presented. The method measures the acoustic amplitude modulation of a small high-frequency signal by a very low-frequency signal. The low-frequency signal is increased to produce a target peak-to-peak modulation. A physical measurement of the peak-to-peak diaphragm excursion is made and defined to be twice Xmax.

Delayed arrivals of a stereo source are manipulated to suppress undesirable vehicle cabin acoustics and replace them with optimum acoustics for reproduction of commercial recordings. These optimum playback acoustics are derived from concert hall and listening room measurements. An experimental vehicle sound system implementing these concepts is described.

Vehicle operation noise, even in the quietest cars, produces high sound pressure levels (SPL) at very low frequencies. This noise masks desired signals in and above this frequency range. A blind subjective test, using ten listeners, was undertaken to determine a frequency response equalization curve that would compensate for this noise under specific but realistic conditions. Starting with a 4 dB full-band level increase, an average of 40 listener responses showed a gentle rise in bass reaching an additional 4 dB at 50 Hz.

A body of listening test techniques which produces consistant rankings of sound systems is presented. Sufficient detail is generated in the course of the prescribed listening evaluation to direct engineering changes to the system. A two dimensional weighting system (performance and usage) is used to determine a single-number rating.

Low-frequency range extension of loudspeakers by alignments which include passive electrical components was first studied by Benson and more recently by Von Recklinghausen. The present work uses a computer model to test the practicality of representative designs. A third-order alignment is found to have considerable merit. Measurements of real systems are presented.

The consumer's listening test is the ultimate test of a sound system. During product development, listening tests must be carefully controlled if they are to produce useful data. Prejudiced or biased listeners are the most frequent cause of poor or misleading data from these tests. This paper describes methods of eliminating the effects of listener bias by using double-blind techniques.