The eternal quest for harmony between measurement and perception
Measurements of electrical and acoustic parameters are essential during the development of speaker systems. They provide the predictable correlations between components, the system design of the signal transmission, and the effect on the signal at the corresponding measurement point. There is an enormous variety of measurement methods and measurement signals in electroacoustics. The most well-known measurement signals are pure sinusoidal tones and noise signals containing a defined mix of frequencies. The range of methods includes recording electrical quantities such as current, voltage and resistance/ impedance, mechanical measurement recordings for vibration analysis, through to the well-known acoustic microphone recordings of radiated signals in a specific spatial arrangement.
Example: Measuring frequency response of sound pressure level
A typical example for an acoustic measurement is the recording of the frequency-related sound pressure level, i.e., the fidelity of a speaker with reference to an input signal, which generally covers the entire spectrum of audible frequencies between 20 Hz and 20 kHz – approx. 10 octaves. However, this spectrum is in practice never continuously – infinitely – reproducible but only by means of a limited count of discrete sinusoidal metering points.
The example of the above frequency response with 405 metering points may show how difficult it is to draw conclusions from the analysis of this measurement to the euphony of a speaker system. Note the extreme zigzag lines and their almost chaotic jumps in the frequency response, especially under true-to-life listening conditions. Obviously, the “true” frequency response in a generic living room is as individual as the living room itself and eventually unpredictable by laboratory/ open-field measurements. Keep in mind, that the human ear is able to differentiate frequencies by +/- 3 to 5 Hz, i.e., less than 1/100 octave in the mid-range audio spectrum.
Regarding the complexity of the “true” measurement diagram it is common practice to level out a speaker box’s original frequency response floating over 1/6 octave or even 1/3 octave. This is what you usually find as part of the documentation for commercial speaker boxes. On the one hand, this is reasonable to find macroscopic effects in the sound characteristics of a speaker box – both under laboratory and true-to-life conditions – or to design crossover networks and other filters. On the other hand, this practice equalizes the apparent sound quality of different speaker boxes to a significant extent.
And still, measurements such as these are very sensitive when it comes to the measurement conditions that are present at that moment, in particular the listening environment. The exact reproduction of an existing measurement recording in a different listening environment remains practically impossible.
Measuring speaker boxes from different positions in a living room.
The above plan view of our sample measurement setup is highly simplified for the sake of visual understanding. Each measurement could change significantly even by small alterations to the interior or architecture, such as repositioning speakers and/ or the couch, adding/removing furniture or by making changes to window areas.
The 3BE approach
During the development process of 3BE’s Aural Sculptures, an in-depth and lengthy detailed development process “in the living room” is connected with the basic electroacoustic engineering. Different people with different listening habits listen to the units we’ve developed for many hours and provide us with feedback. The design is changed often only very slightly until we see the desired glow in the eyes of our test listeners, and there are no longer any “buts” to cloud their judgment. All the listening sessions are carefully documented and build a non-deterministic knowledge base that helps us also for current and future developments.
Believe it or not: Experience has taught us that the optimum sound of a speaker system – particularly under real-life listening conditions – is not necessarily able to be explained by the normative optimum of all physical measures. And one other thing we’ve learned at 3BE: EVERYBODY has a sense of the finer differences in the sound of music reproduction, even if people do not initially consider themselves to be all that sensitive. After all, there is only one with the perfect hearing: