5D Library › Architectural Acoustics, Reverberation Time and the Environmental Psychology of Built Space · 中文
Chapter 8 Practice and a Pre-Registered Trial
Note
Please reread the health and safety notice at the start of the preface. The checklist in this chapter is a tool for recording and thinking, not medical advice and not a building, fire or gas safety inspection. Persistent tinnitus, hearing loss or ear pain, long-standing insomnia or excessive daytime sleepiness call for a doctor. A CO2 reading is no substitute for a carbon monoxide alarm; if a room smells mouldy, pungent or of gas, have it checked by a professional. The lighting targets in this chapter apply only to healthy adults and are not light therapy.
Part 1 A four-step checklist of environmental physics
This checklist turns the measurements of the first six chapters into four steps anyone can take in a study, an office or a meditation hall. It needs no expensive instruments: a phone that can record sound, a sound level meter with A and C weighting (or a calibrated phone app), a CO2 monitor and a lux meter.
- Clap test and soft furnishings. Clap hard once in the middle of the room (or burst a balloon), record it with a phone, look at the energy decay in software, fit a straight line from −5 dB to −25 dB using Schroeder’s backward integration and extrapolate to the reverberation time (Schroeder 1965). For rooms for speech and focused work the target is no more than about 0.6 s; turn the shortfall into the extra absorption needed using Sabine’s formula from Chapter 1, and add it with bookshelves, curtains, rugs or absorbent panels. Note that phone recordings have limited dynamic range and results below 200 Hz are unreliable (Chapter 2). (Chapters 1 and 2)
- STI isolation from background speech in the work area. At the desk, measure the A-weighted level of a neighbour speaking normally and the background level when no one speaks, and estimate the STI; the target is an STI below 0.50 for the neighbour’s speech. Remedies, in order: increase distance or change seats, add screens and absorption, designate quiet zones, and only then masking sound; Chapter 3 showed that masking does almost nothing for neighbours within 1–2 m. (Chapters 1 and 3)
- Meeting-room CO2 and a ventilation calculation. When people leave after a meeting, note the CO2 reading, note it again an hour later and work out the air change rate from the decay; then use the formula of Chapter 6 for the steady-state concentration with the usual number of people. This book chooses 800 ppm as a conservative design target for meeting rooms (a design choice, not a health threshold). (Chapter 6)
- Desk view and daytime light audit. Record what can be seen from the desk through the window (building outlines or trees and ridges) and, if wished, measure its box-counting dimension by a fixed method; measure the vertical illuminance at the eye once in the afternoon and once in the evening, multiply by the melanopic DER of the light source and check whether it reaches 250 lux in the day and stays at or below 10 lux in the evening (Chapter 5). Where the view cannot be changed, plants can be placed in the line of sight; but Chapter 4 showed that any specific target dimension is only a hypothesis. (Chapters 4 and 5)
Calculation
The checklist applied to a fictional study R (4.0 × 3.5 × 2.7 m, used by two people). Step 1 uses a synthetic clap recording made by the program (true reverberation time 0.9 s, with a noise floor); all other readings are invented.
Calculation
# Four-step environment checklist on a FICTIONAL study room R (4.0 x 3.5 x 2.7 m, two
# people). Step 1 uses a synthetic clap recording (fixed seed); the other inputs are
# invented readings.
import math, random
V = 4.0 * 3.5 * 2.7
# step 1: clap test -> Schroeder backward integration, fit -5 to -25 dB (T20) -> RT60
rng, FS, T_TRUE = random.Random(31), 8000, 0.9
h = [rng.gauss(0, 1) * math.exp(-6.91 * i / FS / T_TRUE) + rng.gauss(0, 10 ** (-55 / 20))
for i in range(int(1.6 * FS))]
energy = [x * x for x in h[: int(1.2 * FS)]] # stop before the noise floor
tail, s = [], 0.0
for e in reversed(energy):
s += e; tail.append(s)
tail.reverse()
db = [10 * math.log10(t / tail[0]) for t in tail]
i5 = next(i for i, d in enumerate(db) if d <= -5)
i25 = next(i for i, d in enumerate(db) if d <= -25)
xs, ys = [i / FS for i in range(i5, i25)], db[i5:i25]
mx, my = sum(xs) / len(xs), sum(ys) / len(ys)
slope = sum((x - mx) * (y - my) for x, y in zip(xs, ys)) / sum((x - mx) ** 2 for x in xs)
T_est = -60 / slope
add = 0.161 * V / 0.5 - 0.161 * V / T_est
print(f"step 1 clap test: RT60 (from T20) {T_est:.2f} s (synthetic room: {T_TRUE} s)")
print(f" to reach 0.5 s add {add:.1f} m2 of absorption,"
f" e.g. {add / 0.8:.1f} m2 of panels with alpha 0.8")
# step 2: neighbour's speech at the desk; STI from SNR with T 0.5 s (simplified MTF as
# in p02)
def sti(T, snr):
tis = []
for F in [10 ** (k / 10) for k in range(-2, 12)]:
m = 1 / math.sqrt(1 + (2 * math.pi * F * T / 13.8) ** 2) / (1 + 10 ** (-snr / 10))
tis.append((max(-15.0, min(15.0, 10 * math.log10(m / (1 - m)))) + 15) / 30)
return sum(tis) / len(tis)
for label, speech, bg in (("as found", 52, 38), ("screen -6 dB", 46, 38),
("screen + masking 44 dB(A)", 46, 44)):
s = sti(0.5, speech - bg)
print(f"step 2 {label:25s}: speech {speech}, bg {bg} dB(A) -> STI {s:.2f}"
+ (" (aim < 0.50)" if s >= 0.5 else " ok"))
# step 3: CO2 decay after the room empties -> air change rate; then the 2-person steady
# state
c_start, c_after, hours, c_out = 1250.0, 700.0, 1.0, 420.0
ach = math.log((c_start - c_out) / (c_after - c_out)) / hours
css = c_out + 2 * 0.005e-3 * 3600 * 1e6 / (ach * V)
need = 2 * 0.005 / ((800 - c_out) * 1e-6) * 3.6 / V
print(f"step 3 CO2 decay 1250 -> 700 ppm in 1 h: ACH {ach:.2f} /h; two people -> steady"
f" {css:.0f} ppm")
print(f" staying below 800 ppm with two people needs ACH {need:.1f} /h")
# step 4: view and light
print("step 4 desk view: block skyline, D about 1.0 (p08) -> add plants or trees")
for when, lux, der, limit, op in (("14:00 at the eye", 180, 0.5, 250, ">="),
("21:30 at the eye", 60, 0.45, 10, "<=")):
medi = lux * der
ok = medi >= limit if op == ">=" else medi <= limit
print(f" {when}: {lux} lx x DER {der} = {medi:.0f} lx mEDI"
f" (target {op} {limit})"
+ (" ok" if ok else " -> change"))
# step 1 clap test: RT60 (from T20) 0.91 s (synthetic room: 0.9 s)
# to reach 0.5 s add 5.5 m2 of absorption, e.g. 6.8 m2 of panels with alpha 0.8
# step 2 as found : speech 52, bg 38 dB(A) -> STI 0.69 (aim < 0.50)
# step 2 screen -6 dB : speech 46, bg 38 dB(A) -> STI 0.61 (aim < 0.50)
# step 2 screen + masking 44 dB(A): speech 46, bg 44 dB(A) -> STI 0.48 ok
# step 3 CO2 decay 1250 -> 700 ppm in 1 h: ACH 1.09 /h; two people -> steady 1296 ppm
# staying below 800 ppm with two people needs ACH 2.5 /h
# step 4 desk view: block skyline, D about 1.0 (p08) -> add plants or trees
# 14:00 at the eye: 180 lx x DER 0.5 = 90 lx mEDI (target >= 250) -> change
# 21:30 at the eye: 60 lx x DER 0.45 = 27 lx mEDI (target <= 10) -> changeStep 1 estimates 0.91 s from the synthetic recording (true value 0.9 s); bringing it down to 0.5 s needs about 5.5 m² of additional equivalent absorption, for example about 6.8 m² of panels with an absorption coefficient of 0.8; a full bookshelf on one wall plus a thick rug comes close to this amount. Step 2: with the neighbour’s speech at 52 dB(A) and a background of 38 dB(A), the STI is 0.69; screens that lower the speech by 6 dB give 0.61, still above the target; only adding masking at 44 dB(A) brings it to 0.48. Step 3: CO2 falls from 1,250 to 700 ppm in an hour, an air change rate of about 1.09 per hour; with two people using the room for long periods the steady state is about 1,296 ppm, and staying below 800 ppm needs about 2.5 air changes per hour, that is, opening windows or adding mechanical ventilation. Step 4: the window looks onto box-like buildings; in the afternoon, 180 lux at the eye with a DER of 0.5 gives only 90 lux melanopic EDI, below 250; at half past nine in the evening, 60 lux with a DER of 0.45 gives 27 lux, above 10, so the lights should be dimmed or replaced with warmer, dimmer ones.
Note
A fictional record is not evidence. The numbers above only show how the checklist calculates and flags. Its value lies in writing judgements down so that they can be checked later, not in computing the “right” room. It neither involves nor admits any claim about auspicious or inauspicious orientations: Chapter 7 explained that this book discusses only environmental variables that can be tested by acoustic, optical and ventilation measurements.
Part 2 Protocol for a pre-registered randomised crossover trial
Question
At the same background level of 55 dB(A), does masking a recording of intelligible conversation (STI about 0.65) with pink noise and water sound until it is unintelligible (STI below 0.25) reduce adults’ error rate in a visually presented serial-recall task?
Chapter 3 explains why the level is held fixed: the irrelevant speech effect is almost independent of level, and office masking precisely adds sound. If masking works, its effect should come from reducing the intelligibility and spectral change of the background speech, not from “more quiet”.
Design
| Element | Specification |
|---|---|
| Design | Within-participant randomised crossover (AB/BA); two test sessions 2–7 days apart at the same time of day; one acoustic condition per session |
| Participants | Adults aged 18–60 whose first language is the language of the recorded conversation; pure-tone screening at 250–8,000 Hz no worse than 20 dB HL; no recruitment from the author’s community, students, customers or card users, and no targeted recruitment of readers of websites run by the author |
| Exclusions | Hearing aids or cochlear implants, persistent tinnitus, medication affecting attention or memory, known neurological conditions affecting cognition, severe lack of sleep in the past week (self-reported) |
| Randomisation | 1:1:1:1 to four sequences (condition order AB or BA × material order XY or YX), blocks of 4, so that AB and BA each receive half; generated and concealed by an independent data centre and revealed only at the start of the first session |
| Condition A (speech) | A recorded everyday conversation between two talkers, played over loudspeakers at 55 dB(A) at the listening position; STI measured in the test room about 0.65 |
| Condition B (masked) | The same recording with a mixed masker of pink noise and water sound, total level still 55 dB(A), measured STI below 0.25 (by the simplified formula of Chapter 1 the masker must be about 7–8 dB above the speech; measurement decides) |
| Parallel versions | Two sets (X, Y) of serial-recall and reading materials matched for difficulty, balanced against condition and order through the four sequences above, so that every condition × version × order combination has the same number of participants (18 per sequence when 72 are randomised) |
| Blinding | Participants hear the sound and cannot be blinded; they are not told the hypothesis (only that the study “examines the effect of background sound on memory”); the session leader does not score; scoring is automatic; the statistician sees only conditions coded 1 and 2 |
| Primary outcome (single) | Serial-recall error rate: in each session, 30 lists of 9 digits presented visually one at a time and recalled in order, scored as the percentage of digits not recalled in the correct position |
| Secondary outcomes | Reading-comprehension accuracy; NASA-TLX workload; rated disturbance; acceptability of the background sound |
| Manipulation check | STI and A-weighted level of both conditions measured in the test room before the first participant, then weekly |
| Analysis | Linear mixed model on the error rate of each session, with fixed effects for condition, period (first or second session) and version and a random effect for participant; the estimand is the condition effect (A − B, speech minus masked) with its 95 % CI; intention-to-treat; sensitivity analysis using first-period data only (between-group comparison) |
| Safety | About an hour at 55 dB(A) is far below levels that risk hearing (occupational hearing-conservation action levels are commonly an 8-hour equivalent of 85 dB(A)); participants may stop at any time; testing ends at once if they feel unwell |
Note
Order and carry-over effects. Serial recall shows clear practice effects, so scores are usually better in the second session; the AB/BA design places each condition in the first and second session equally often, so the period effect can be estimated and removed in the model. The effect of background sound is immediate, and a gap of several days is ample to remove any physiological carry-over, but participants may learn some “strategy for coping with background sound” in their first session, and such strategic carry-over cannot be ruled out entirely. The protocol does not pre-test for carry-over (such pre-tests themselves distort inference); instead it pre-specifies a sensitivity analysis using first-period data only, and if the two analyses point in different directions the report must present both. The report follows the CONSORT extension for randomised crossover trials (Dwan et al. 2019).
Sample size
Calculation
# Sample size for the AB/BA crossover: primary outcome = serial-recall error rate (% of
# items), within-person difference D = errors(intelligible speech) - errors(masked
# speech), both 55 dB(A). Paired normal approximation with the usual small-sample
# correction z_a^2 / 2; rounded up to an even number so that both orders are equally
# filled. Assumptions (illustrative, no direct published estimate for this contrast):
# smallest effect of interest 5 pp, SD of D 12 pp.
import math
from statistics import NormalDist
z = NormalDist().inv_cdf
def n_pairs(delta, sd, power=0.90, alpha=0.05):
za = z(1 - alpha / 2)
n = (za + z(power)) ** 2 * sd ** 2 / delta ** 2 + za ** 2 / 2
n = math.ceil(n)
return n + n % 2
print(" SESOI (pp) | SD of D = 10 | 12 | 15 (analysable, 90 % power, alpha 0.05)")
for d in (4, 5, 6, 8):
a, b, c = n_pairs(d, 10), n_pairs(d, 12), n_pairs(d, 15)
print(f" {d:2d} | {a:4d} | {b:3d} | {c:3d}")
n = n_pairs(5, 12)
recruit = math.ceil(n / 0.90); recruit += recruit % 2
print(f"chosen: 5 pp, SD 12 pp -> {n} analysable"
f" (80 % power: {n_pairs(5, 12, power=0.80)})")
print(f"allowing 10 % loss: randomise {recruit} ({recruit // 2} to each order AB and BA)")
se = 12 / math.sqrt(n); P = NormalDist().cdf
print(f"SE of the mean difference {se:.2f} pp; if the true effect is 5 pp:")
print(f" P(lower 95 % limit > 0) = {P(5 / se - 1.96):.0%},"
f" P(lower limit >= 5) = {P(-1.96):.1%}")
print(f"a 90 % chance that the lower limit reaches 5 needs a true effect of"
f" {5 + (1.96 + z(0.90)) * se:.1f} pp")
# SESOI (pp) | SD of D = 10 | 12 | 15 (analysable, 90 % power, alpha 0.05)
# 4 | 68 | 98 | 150
# 5 | 44 | 64 | 98
# 6 | 32 | 44 | 68
# 8 | 20 | 26 | 40
# chosen: 5 pp, SD 12 pp -> 64 analysable (80 % power: 48)
# allowing 10 % loss: randomise 72 (36 to each order AB and BA)
# SE of the mean difference 1.50 pp; if the true effect is 5 pp:
# P(lower 95 % limit > 0) = 92%, P(lower limit >= 5) = 2.5%
# a 90 % chance that the lower limit reaches 5 needs a true effect of 9.9 ppThe smallest effect of interest is set at 5 percentage points. This is below the 8–14 points of the illustrative calculation in Chapter 3, a conservative choice: an effect worth investing in an office masking system for should be at least this large. The standard deviation of the within-participant differences is set at 12 points (illustrative; individual differences in the irrelevant speech effect are large, and we found no published estimate for this specific contrast). With two-sided and 90 % power, 64 analysable participants are needed; allowing for about 10 % loss, 72 are randomised, 36 to AB and 36 to BA. A standard deviation of 15 points, or a smallest effect of 4 points, would each need 98; the sensitivity table is published at pre-registration.
To be clear: 90 % power means that if the true effect is exactly 5 points, the probability that the lower limit of the 95 % confidence interval lies above 0 is about 92 %; the probability that the lower limit reaches 5, allowing the statement “at least the smallest effect of interest”, is only 2.5 %. For a 90 % chance of the stronger statement, the true effect would need to be about 9.9 points.
Decision rule
Calculation
# Decision rule fixed before enrolment. diff = errors(speech) - errors(masked),
# percentage points, mean of within-person differences adjusted for period (AB/BA), with
# its 95 % CI. SESOI 5 pp. Manipulation check (measured in the test booth before any
# participant): speech condition STI >= 0.60, masked condition STI <= 0.25, A-weighted
# levels within 1 dB of 55 dB(A).
SESOI = 5.0
def verdict(diff, lo, hi, sti_speech=0.65, sti_masked=0.22, level_gap=0.4, complete=0.95,
first_period_sign_agrees=True):
if sti_speech < 0.60 or sti_masked > 0.25 or level_gap > 1.0:
return f"diff {diff:+.1f}: NOT INTERPRETABLE - acoustic manipulation check failed"
if complete < 0.85:
return f"diff {diff:+.1f}: NOT INTERPRETABLE - < 85 % completed both sessions"
if lo >= SESOI: v = "masking lowers errors by at least the SESOI"
elif lo > 0: v = "masking lowers errors; size not shown to reach 5 pp"
elif hi < 0: v = "masking RAISES errors"
elif lo >= -SESOI and hi <= SESOI: v = "no difference worth acting on"
else: v = "unclear: the interval is too wide to decide"
note = "\n [flag: first-period-only estimate disagrees]"
note = "" if first_period_sign_agrees else note
return f"diff {diff:+.1f} pp ({lo:+.1f} to {hi:+.1f}): {v}{note}"
for case in ((9.0, 6.0, 12.0), (6.0, 3.0, 9.0), (1.0, -2.0, 4.0), (-4.0, -7.0, -1.0),
(2.0, -5.5, 9.5), (9.0, 6.0, 12.0, 0.55),
(9.0, 6.0, 12.0, 0.65, 0.22, 0.4, 0.80),
(6.0, 3.0, 9.0, 0.65, 0.22, 0.4, 0.95, False)):
print(verdict(*case))
print("secondary outcomes (reading comprehension, NASA-TLX workload, rated disturbance)")
print("are reported with intervals and never decide")
# diff +9.0 pp (+6.0 to +12.0): masking lowers errors by at least the SESOI
# diff +6.0 pp (+3.0 to +9.0): masking lowers errors; size not shown to reach 5 pp
# diff +1.0 pp (-2.0 to +4.0): no difference worth acting on
# diff -4.0 pp (-7.0 to -1.0): masking RAISES errors
# diff +2.0 pp (-5.5 to +9.5): unclear: the interval is too wide to decide
# diff +9.0: NOT INTERPRETABLE - acoustic manipulation check failed
# diff +9.0: NOT INTERPRETABLE - < 85 % completed both sessions
# diff +6.0 pp (+3.0 to +9.0): masking lowers errors; size not shown to reach 5 pp
# [flag: first-period-only estimate disagrees]
# secondary outcomes (reading comprehension, NASA-TLX workload, rated disturbance)
# are reported with intervals and never decideThe example results are invented. Only if the lower limit of the 95 % confidence interval of the effect (error rate under speech minus under masking) is at least 5 percentage points will the report say “masking reduced serial-recall errors by at least the smallest effect of interest”. If the manipulation check fails (speech-condition STI below 0.60, masked-condition STI above 0.25, or levels of the two conditions differing by more than 1 dB), or fewer than 85 % of participants complete both sessions, the report says “not interpretable” whatever the result. If the estimate from first-period data alone points in a different direction from the main analysis, the report adds a flag. Secondary outcomes are reported only as estimates with intervals and play no part in the decision.
What this trial cannot show
First, it compares two background sounds and includes no quiet condition, so it cannot estimate the total disruption caused by “speech itself”, only how much of it masking recovers. Second, masking lowers both STI and spectral change, so the trial cannot distinguish the two mechanisms of Chapter 3; that would need an additional “foreign language not understood” condition. Third, a recording is not a living colleague: in a real office, talkers look at you and mention people you know, and the disruption may be larger. Fourth, a test of about an hour says nothing about long-term adaptation, satisfaction or productivity. Fifth, participants are adults with normal hearing who speak the recording’s language as their first language; the results cannot be extended to children, people with hearing loss or non-native listeners. Sixth, it can lend no support to any “fengshui layout” or any product: the trial tests one acoustic variable.
Ethics and conflicts of interest
The trial must be approved by an independent ethics committee, participants must give informed consent, and the protocol and analysis plan are registered before the first participant is enrolled (Nosek et al. 2018). Participants may withdraw at any time without giving a reason. The results will be published whether positive or negative.
Disclosure
The author uses the works of 5D Library, including this book, as the methodological basis of his community courses, the 5D Ascension Cards (Dimensional Shift Card Decks) and the Yafu AI knowledge base, and therefore has a commercial interest in the subject of this book. In addition, the author operates websites on spatial aesthetics (such as 5dhouse.com) and therefore has a commercial interest in the subject of this book; the book names no brand or product and recommends no renovation, fengshui or interior-design service. The book promotes no course, card deck or service; the Chapter 8 trial is designed to be run and analysed by researchers independent of the author’s business and does not recruit from the author’s community, students or customers; all the code and the study protocol are published openly for anyone to check.
To limit the influence of these interests, the trial materials, analysis code and decision rule are published before enrolment, and the results, data and code will be published whatever the outcome.
References
Fan, Y. (2026). Architectural Acoustics, Reverberation Time and the Environmental Psychology of Built Space: From Sabine's Equation and Fractal Visual Dimensions to Cognitive Performance, Read Beside Classical Chinese Spatial Design [建筑声学、混响时间与空间环境心理学的物理模型] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23782090 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).