5D Library › Rhythmic Acoustic Stimulation, Auditory Steady-State Responses and Modulation Signal Engineering · 中文
Chapter 4. Delivery Routes and Gating Shapes
Leakage turns a binaural beat into a monaural one
Chapter 1 showed that a binaural beat needs each tone to reach only its own cochlea. Real delivery always leaks: the right-ear tone reaches the left cochlea with some interaural attenuation (IA). With leaked amplitude ratio the left ear hears , whose amplitude envelope swings between and ; so the monaural beat has amplitude modulation depth exactly , and the power envelope has depth .
Calculation
# Chapter 4. p07: When does a binaural beat turn into a monaural one?
# If the right-ear tone reaches the left cochlea with amplitude ratio
# a = 10^(-IA / 20), where IA is the interaural attenuation in dB, the
# left ear hears cos(w t) + a cos((w + dw) t). Its amplitude envelope
# |1 + a exp(i dw t)| swings between 1 - a and 1 + a, so the monaural
# beat depth is exactly a, and the power envelope depth is
# 2a / (1 + a^2). The IA values below are typical orders of magnitude
# for each way of delivering sound, used for illustration only.
# The program also checks the formula against a direct computation of
# the envelope on a fine time grid.
import math
DEVICES = [
("insert earphones", 70.0),
("closed circumaural", 50.0),
("supra-aural on-ear", 40.0),
("open-back headphones", 25.0),
("bone conduction band", 5.0),
("stereo loudspeakers", 3.0),
]
def direct_depth(a, n=20000):
env = [abs(1 + a * complex(math.cos(2 * math.pi * k / n),
math.sin(2 * math.pi * k / n)))
for k in range(n)]
return (max(env) - min(env)) / (max(env) + min(env))
if __name__ == "__main__":
print(" route IA dB leak a amp depth"
" power depth")
for name, ia in DEVICES:
a = 10 ** (-ia / 20)
print(" {:24s} {:6.0f} {:8.4f} {:10.4f} {:12.4f}".format(
name, ia, a, a, 2 * a / (1 + a * a)))
print("\n check against direct envelope computation:")
for a in (0.01, 0.1, 0.5):
print(" a = {:4.2f} formula {:.4f} direct {:.4f}".format(
a, a, direct_depth(a)))
print("\nReading: through insert or closed headphones the leaked")
print("beat is a fraction of a percent deep and the binaural")
print("mechanism is all there is. A bone conduction band or a pair")
print("of loudspeakers delivers both tones to both cochleae, so a")
print("'binaural' track becomes an ordinary monaural beat of more")
print("than half depth, which acoustically is close to an")
print("isochronic tone. Whatever such a track does, it is not")
print("doing it binaurally.")
# route IA dB leak a amp depth power depth
# insert earphones 70 0.0003 0.0003 0.0006
# closed circumaural 50 0.0032 0.0032 0.0063
# supra-aural on-ear 40 0.0100 0.0100 0.0200
# open-back headphones 25 0.0562 0.0562 0.1121
# bone conduction band 5 0.5623 0.5623 0.8545
# stereo loudspeakers 3 0.7079 0.7079 0.9432
#
# check against direct envelope computation:
# a = 0.01 formula 0.0100 direct 0.0100
# a = 0.10 formula 0.1000 direct 0.1000
# a = 0.50 formula 0.5000 direct 0.5000
#
# Reading: through insert or closed headphones the leaked
# beat is a fraction of a percent deep and the binaural
# mechanism is all there is. A bone conduction band or a pair
# of loudspeakers delivers both tones to both cochleae, so a
# 'binaural' track becomes an ordinary monaural beat of more
# than half depth, which acoustically is close to an
# isochronic tone. Whatever such a track does, it is not
# doing it binaurally.The IA of each route is an illustrative order of magnitude. Through insert earphones the leaked monaural beat is only 0.0003 deep, through supra-aural on-ear headphones 0.0100, through open-back headphones 0.0562; in these cases the binaural mechanism is the only one. A bone conduction band carries the vibration to both cochleae, giving depth 0.5623; stereo loudspeakers give 0.7079, with a power envelope 0.9432 deep. A “binaural beat” played through loudspeakers or bone conduction is therefore acoustically close to an isochronic tone. Whatever such a track does, it is not doing it binaurally. The program also checks the formula against a directly computed envelope; the three leakage ratios agree.
Gating an isochronic tone
An isochronic tone switches a carrier on and off at a fixed rate. The edges can be abrupt, a square gate, or smoothed by raised-cosine ramps of a few milliseconds. Abrupt edges throw part of the power far from the carrier, heard as a click at each onset; ramps change the shape of the envelope, and so the strength of the spectral line at the stimulation rate and the crest factor of the waveform.
Calculation
# Chapter 4. p08: Gating shape of an isochronic tone: clicks and peaks.
# An isochronic tone switches a 400 Hz carrier on and off at 10 Hz with
# a 50 % duty cycle. The gate edges can be abrupt (square), or ramped
# by raised-cosine edges of 5 ms or 20 ms; a sinusoidal amplitude
# modulation (m = 1) is the smoothest case. For one second at 8,192 Hz
# we report:
# splatter: share of power more than 100 Hz away from the carrier,
# heard as clicks; in dB relative to total power;
# crest factor: peak over RMS amplitude, in dB;
# rate line: power at carrier +/- 10 Hz relative to the carrier.
# Splatter below -120 dB is printed as -120 (numerical floor).
import cmath
import math
FS = N = 8192
def fft(x):
n = len(x)
a = list(x)
j = 0
for i in range(1, n):
bit = n >> 1
while j & bit:
j ^= bit
bit >>= 1
j |= bit
if i < j:
a[i], a[j] = a[j], a[i]
size = 2
while size <= n:
w = [cmath.exp(-2j * math.pi * k / size) for k in range(size // 2)]
for s in range(0, n, size):
for k in range(size // 2):
u, v = a[s + k], a[s + k + size // 2] * w[k]
a[s + k], a[s + k + size // 2] = u + v, u - v
size *= 2
return a
def gate(t, ramp):
ph = (t * 10.0) % 1.0
on = 0.5
if ph >= on:
return 0.0
tt = ph / 10.0
if ramp <= 0:
return 1.0
if tt < ramp:
return 0.5 - 0.5 * math.cos(math.pi * tt / ramp)
if tt > on / 10.0 - ramp:
return 0.5 - 0.5 * math.cos(math.pi * (on / 10.0 - tt) / ramp)
return 1.0
def measure(env):
x = [e * math.cos(2 * math.pi * 400 * i / FS)
for i, e in enumerate(env)]
X = fft(x)
p = [abs(v) ** 2 for v in X[:N // 2]]
tot = sum(p)
far = sum(v for k, v in enumerate(p) if abs(k - 400) > 100)
rms = math.sqrt(sum(v * v for v in x) / N)
crest = 20 * math.log10(max(abs(v) for v in x) / rms)
side = 10 * math.log10((p[390] + p[410]) / 2 / p[400])
return max(10 * math.log10(far / tot), -120.0), crest, side
if __name__ == "__main__":
cases = [("square gate", [gate(i / FS, 0) for i in range(N)]),
("5 ms raised cosine", [gate(i / FS, 0.005)
for i in range(N)]),
("20 ms raised cosine", [gate(i / FS, 0.020)
for i in range(N)]),
("sinusoidal AM, m = 1",
[0.5 + 0.5 * math.cos(2 * math.pi * 10 * i / FS)
for i in range(N)])]
print(" gate shape splatter dB crest dB rate line dB")
for name, env in cases:
sp, cr, sd = measure(env)
print(" {:22s} {:10.1f} {:10.2f} {:12.2f}".format(name, sp, cr, sd))
print("\nReading: an abrupt gate throws about 2 % of the power far")
print("from the carrier, heard as a click at every onset. A 20 ms")
print("raised cosine pushes that below -48 dB and, at this duty")
print("cycle, even strengthens the 10 Hz line relative to the")
print("carrier. Sinusoidal modulation is the cleanest spectrum of")
print("all but has the weakest rate line of the four.")
# gate shape splatter dB crest dB rate line dB
# square gate -16.9 6.02 -3.92
# 5 ms raised cosine -24.7 6.60 -3.14
# 20 ms raised cosine -48.3 9.03 -1.65
# sinusoidal AM, m = 1 -120.0 7.27 -6.02
#
# Reading: an abrupt gate throws about 2 % of the power far
# from the carrier, heard as a click at every onset. A 20 ms
# raised cosine pushes that below -48 dB and, at this duty
# cycle, even strengthens the 10 Hz line relative to the
# carrier. Sinusoidal modulation is the cleanest spectrum of
# all but has the weakest rate line of the four.A square gate throws about 2% of the power (16.9 dB down) more than 100 Hz away from the carrier; a 5 ms raised-cosine ramp pushes that share below 24.7 dB down, a 20 ms ramp below 48.3 dB down. At a 50% duty cycle the 20 ms ramp also raises the 10 Hz line from 3.92 dB to 1.65 dB below the carrier. Sinusoidal modulation has the cleanest spectrum but the weakest rhythm line, 6.02 dB below the carrier. Crest factors lie between 6.02 dB and 9.03 dB, 7.27 dB for sinusoidal modulation; this sets how high the peaks are for the same average loudness.
Note
Design consequence. To evoke a steady-state response with an isochronic tone, raised-cosine ramps of a few to twenty milliseconds cost almost nothing and remove the clicks. To compare binaural beats with isochronic tones, the carriers, the average levels and the strength of the rhythm line should be matched in advance; otherwise the comparison is of acoustic parameters, not of mechanisms.
Summary
The delivery route decides whether a binaural beat is still binaural: only headphones with little leakage keep the binaural mechanism, while loudspeakers and bone conduction turn it into a monaural beat. Isochronic gating should use short raised-cosine ramps, and comparisons between stimuli should match the strength of the rhythm line.
Fan, Y. (2026). Rhythmic Acoustic Stimulation, Auditory Steady-State Responses and Modulation Signal Engineering: Binaural versus Isochronic Envelope Depth, 40-Hz Phase-Locking Bounds and Pink-Noise Spectral Masking [节律声刺激、听觉稳态反应与双耳等时调制工程] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090211 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).