Okay, I’m going a little psycho (acoustic). The entire premise for this project is our ear-brain integration time and how it interacts with the acoustic source and room reflections.Background:
Over the past couple of years I’ve followed posts by respected industry professionals from companies like KEF, Snell, JBL, etc. I pay particular attention to their holistic views on speakers and rooms…the perception of sound in real rooms. Among other things, this obviously includes acoustic reflections from room surfaces.
It seems generally accepted that our ear-brain has a very short integration time at high frequencies…we sense first arrival from the speaker, largely oblivious to room reflections. As frequency decreases, integration time increases progressively until, in woofer territory, we fully sense contribution of the room…our ear-brain acts sorta like windowed MLS measurements. I have been unable to find quantitative documentation of this “sliding scale”, but the tendencies seem clear.
Floyd Toole indicates we can detect acoustic flaws as narrow as 1/20 octave. SoundEasy doesn’t have a selection for 1/20 but it does do 1/24 octave which, unless otherwise stated, I’ll use consistently in this thread. Also for consistency, only Hanning windows are shown. (Hanning and other tapered windows like BH and KB emphasize the leading edge of the impulse, trailing off to zero influence at the end of the time window; a rectangular window weights all input equally.)
All measurements shown here are from the 4M listening distance. Why 4M? Many speakers are designed with practically anechoic measurements (1M, gated MLS, GP, etc). This is appropriate when the listening environment is unknown or unpredictable. However, IMO, it is somewhat shortsighted to ignore a known
environment, like our purpose-specific listening rooms. In other words, you will not see the issue discussed here in your system unless you look for it with LP measurements in your room.
EQ of the “old” speakers evolved through hundreds of measurements. The most pleasing results were obtained with “mental integration” of measurements at 2ms, 4, 8, 16, etc, doubling through 128ms. 2ms provides the reference up around 10kHz (including direct arrival only) with increasing room reflections added until 128ms provided the subwoofer reference. Here is an overlaid example to help illustrate the process.
Through the measurement/EQ process, floor reflection impact was obvious, even with short “listening” windows. For example, a 2ms Hanning window timed to exclude floor bounce yields clean first arrival frequency response. Including floor bounce by extending even the very forgiving tapered window to 3ms begins to corrupt FR. To illustrate rapid onset of the issue, here is one of the “old” speakers from 4M, at 2ms, then 3, 5, 10, 20, and finally 100ms. (BTW, this and all following MLS FR plots in this post are run at the same level but manually separated for clarity.) The longer windows allow other reflections to enter the picture, but floor bounce still dominates midrange corruption because it is the nearest surface with the strongest return. Also notice the apparent increase in bass levels as longer integration time and the room come onboard.
The Octagon ceiling is already heavily treated with fiberglass stuffed in the space between (fabric covered) joists, but ten 2’x2’ floor pillows were the only practical tool I had to combat floor bounce. Unfortunately the pillows are not tunable and positioning them for improvement usually met with degraded performance elsewhere. No pillows were used in any of the measurements in this thread.