5D Library › 节律声刺激、听觉稳态反应与双耳等时调制工程 · English
第一章 包络在哪里:双耳拍与等时音
三种「拍」
两个频率相近的纯音同时作用于听觉系统,可以产生三种不同的东西。单耳拍:两个音在同一只耳朵里(或在空气中)相加,合成波形的振幅以两者频差为周期起伏,这是一种物理上存在的包络。双耳拍:左耳只听到 ,右耳只听到 ,两只耳朵各自收到的都是振幅恒定的纯音;听者仍然感到以 起伏的拍,但这一感觉产生于脑干对两耳相位的比较 (Oster 1973)。等时音:一个载波被直接开关或调幅,节律写在波形本身里。
三者的区别可以用一个量来刻画:到达每只耳朵的信号包络的调制深度。包络取解析信号的模,即希尔伯特包络;调制深度定义为包络的(最大值减最小值)除以(最大值加最小值)。
推导
# Chapter 1. p01: What envelope reaches each ear? Binaural vs isochronic.
# One second of sound sampled at 8,192 Hz. The envelope of a signal is
# the magnitude of its analytic signal, obtained by a discrete Fourier
# transform with the negative frequencies removed (a Hilbert envelope).
# Modulation depth = (max - min) / (max + min) of the envelope, edges
# trimmed. Cases (carrier f0 = 400 Hz, beat or rate df = 10 Hz):
# L ear of a binaural pair: cos(2 pi f0 t) alone;
# R ear of a binaural pair: cos(2 pi (f0 + df) t) alone;
# both tones mixed in the air (loudspeakers) or by bone conduction;
# L ear through headphones with 40 dB interaural leakage;
# isochronic tone [1 + m cos(2 pi df t)] cos(2 pi f0 t), m = 1, 0.5;
# a static 432 Hz tone with no modulation at all.
import cmath
import math
FS = N = 8192
def fft(x):
n = len(x)
if n == 1:
return [x[0]]
ev, od = fft(x[0::2]), fft(x[1::2])
out = [0j] * n
for k in range(n // 2):
w = cmath.exp(-2j * math.pi * k / n) * od[k]
out[k], out[k + n // 2] = ev[k] + w, ev[k] - w
return out
def ifft(x):
n = len(x)
y = fft([v.conjugate() for v in x])
return [v.conjugate() / n for v in y]
def envelope(x):
X = fft([complex(v) for v in x])
h = [0.0] * N
h[0] = h[N // 2] = 1.0
for k in range(1, N // 2):
h[k] = 2.0
return [abs(v) for v in ifft([a * b for a, b in zip(X, h)])]
def depth(x):
e = envelope(x)[N // 8: -N // 8]
return (max(e) - min(e)) / (max(e) + min(e))
def tone(f, a=1.0):
return [a * math.cos(2 * math.pi * f * i / FS) for i in range(N)]
if __name__ == "__main__":
f0, df = 400.0, 10.0
L, R = tone(f0), tone(f0 + df)
leak = 10 ** (-40 / 20)
cases = [
("binaural, left ear alone", L),
("binaural, right ear alone", R),
("both tones mixed in air", [a + b for a, b in zip(L, R)]),
("left ear, 40 dB leakage", [a + leak * b for a, b in zip(L, R)]),
]
for m in (1.0, 0.5):
cases.append(("isochronic, m = {}".format(m), [
(1 + m * math.cos(2 * math.pi * df * i / FS)) * c
for i, c in enumerate(L)]))
cases.append(("static 432 Hz tone", tone(432.0)))
print(" signal envelope modulation depth")
for name, x in cases:
print(" {:30s} {:12.4f}".format(name, depth(x)))
print("\nReading: each ear of a binaural pair receives a steady")
print("tone with no envelope at all; the beat exists only after the")
print("two ears are compared in the brainstem. Played through the")
print("air, the two tones mix and the beat becomes an ordinary")
print("monaural one of full depth. An isochronic tone carries its")
print("rhythm in the waveform itself; a static tone carries none.")
# signal envelope modulation depth
# binaural, left ear alone 0.0000
# binaural, right ear alone 0.0000
# both tones mixed in air 1.0000
# left ear, 40 dB leakage 0.0100
# isochronic, m = 1.0 1.0000
# isochronic, m = 0.5 0.5000
# static 432 Hz tone 0.0000
#
# Reading: each ear of a binaural pair receives a steady
# tone with no envelope at all; the beat exists only after the
# two ears are compared in the brainstem. Played through the
# air, the two tones mix and the beat becomes an ordinary
# monaural one of full depth. An isochronic tone carries its
# rhythm in the waveform itself; a static tone carries none.结果很干脆。双耳拍的左耳、右耳单独看,调制深度都是 0.0000:耳蜗收到的是没有节律的稳态音。同样两个音在空气中混合,调制深度为 1.0000,这已是一个满深度的单耳拍。耳机若有 40 dB 的两耳间串音,左耳也只得到 0.0100 的包络。等时音的深度就是它的调制指数,1.0000 或 0.5000。静态的 432 Hz 纯音,深度为 0。
双耳拍依赖相位锁定
既然双耳拍的节律不在声波里,它就只能来自听神经对载波细结构的锁相:两侧神经必须分别跟上各自的载波,脑干才有相位可比。听神经的锁相能力随频率升高而迅速下降,通常用向量强度衡量 (Johnson 1980)。双耳拍的经典心理声学研究也表明,可感知的双耳拍只存在于较低的载波上 (Licklider et al. 1950; Perrott and Nelson 1969)。
证据
已发表的证据。 Licklider、Webster 与 Hedlun 以及 Perrott 与 Nelson 测定了双耳拍可被察觉的载波频率与拍频上限 (Licklider et al. 1950; Perrott and Nelson 1969);Johnson 系统测量了猫听神经纤维对单音的同步(锁相)程度随频率的变化 (Johnson 1980)。具体阈值因受试者与测量方法而异,本书只使用其方向:载波越高,锁相越弱。
程序用一条示意的向量强度曲线 表示锁相随频率的衰减,把双耳拍的有效驱动设为 (两侧都必须锁相;0.25 表示双耳拍反应弱于满深度调制),等时音的驱动等于其调制深度 ,然后比较两者的驱动功率。
推导
# Chapter 1. p02: Why binaural beats fade with carrier frequency.
# A binaural beat exists only if both auditory nerves follow the fine
# structure of their tone, so that the brainstem can compare phases.
# Phase locking is measured by the vector strength VS(f) (1 = perfect,
# 0 = none); the curve used here is an illustrative shape, 0.85 below a
# few hundred Hz and falling steeply above 1 kHz:
# VS(f) = 0.85 exp(-(f / 1000)^2).
# The effective central drive of a binaural beat is taken as
# b(f) = 0.25 VS(f)^2 (both ears must lock; 0.25 scales the weaker
# binaural response relative to a fully modulated tone). An isochronic
# tone of depth m drives with m. If the steady-state response grows in
# proportion to the drive, its power ratio is (m / b)^2, in dB.
import math
def vs(f):
return 0.85 * math.exp(-(f / 1000) ** 2)
if __name__ == "__main__":
print(" carrier Hz VS(f) binaural drive isochronic m=1")
print(" advantage (dB)")
for f in (250, 400, 700, 1000, 1500, 2000):
b = 0.25 * vs(f) ** 2
print(" {:10,d} {:7.3f} {:16.5f} {:16.1f}".format(
f, vs(f), b, 20 * math.log10(1.0 / b)))
m = 0.5
b = 0.25 * vs(400) ** 2
print(" isochronic m = 0.5 at 400 Hz still ahead by {:.1f} dB".format(
20 * math.log10(m / b)))
print("\nReading: below about 1 kHz the isochronic tone leads by")
print("one to two orders of magnitude in power; above 1.5 kHz the")
print("nerves no longer follow the carrier and the binaural drive")
print("all but vanishes, so the lead grows by further orders of")
print("magnitude. The exact figures rest on the assumed curve, the")
print("direction does not.")
# carrier Hz VS(f) binaural drive isochronic m=1
# advantage (dB)
# 250 0.799 0.15940 16.0
# 400 0.724 0.13116 17.6
# 700 0.521 0.06779 23.4
# 1,000 0.313 0.02444 32.2
# 1,500 0.090 0.00201 54.0
# 2,000 0.016 0.00006 84.4
# isochronic m = 0.5 at 400 Hz still ahead by 11.6 dB
#
# Reading: below about 1 kHz the isochronic tone leads by
# one to two orders of magnitude in power; above 1.5 kHz the
# nerves no longer follow the carrier and the binaural drive
# all but vanishes, so the lead grows by further orders of
# magnitude. The exact figures rest on the assumed curve, the
# direction does not.在 400 Hz 载波下,向量强度为 0.724,双耳拍驱动为 0.13116,满深度等时音的驱动功率比它高 17.6 dB;即使等时音只调制一半,仍高 11.6 dB。载波升到 1,000 Hz,差距为 32.2 dB;1.5 kHz 以上神经几乎不再跟随载波,差距在 2,000 Hz 时达到 84.4 dB。这些数值依赖假设的曲线,方向不依赖:双耳拍是一种弱而脆弱的刺激,它的载波必须低。
假说 · 尚未验证
可检验的推论。 若双耳拍确实通过脑干相位比较起作用,则在同一受试者身上,把载波从 400 Hz 升到 1,500 Hz 应使拍频处的稳态反应显著减弱,而同频率的等时音不受此影响。什么会证明这一推论错:若高载波双耳拍的反应与低载波相同,则其反应并非来自两耳相位比较,需另寻机制(例如串音形成的单耳拍)。
小结
双耳拍的节律不在任何一只耳朵的声波里,只在脑干里;它依赖锁相,因此只能使用低载波,驱动也远弱于等时音。在空气中播放的「双耳拍」已经变成单耳拍。静态纯音没有包络,在这一层面上无节律可供大脑跟随。
参考文献
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).