Python 3D 频谱图（类似于 matlab 的频谱图函数）

Question

我的问题如下：

我拥有频谱图所需的所有值 (scipy.fftpack.fft)。我想在 python 中创建一个 3D 频谱图。

在 MATLAB 中，这是一个非常简单的任务，而在 python 中，它似乎要复杂得多。我尝试了 mayavi，3D 绘图 matplotlib，但我没有设法做到这一点。

谢谢

我的代码：

import numpy as np
import pandas as pd
import numpy as np
from scipy import signal
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
from matplotlib import cm
from matplotlib.collections import PolyCollection

fs = 11240.
t = 10
time = np.arange(fs*t) / fs
frequency = 1000.
mysignal = np.sin(2.0 * np.pi * frequency * time)

nperseg = 2**14
noverlap = 2**13
f, t, Sxx = signal.spectrogram(mysignal, fs, nperseg=nperseg,noverlap=noverlap)

myfilter = (f>800) & (f<1200)

fig,ax = plt.subplots()

plt.pcolormesh(t, f[myfilter], 10*np.log10(Sxx[myfilter, :]), cmap='jet')
plt.show()

fig = plt.figure()
ax = fig.gca(projection='3d')
x = []
y = []

for counter,i in enumerate(f):
    x.append(np.array([i for k in t]))
    y.append(t)

ax.plot_surface(np.array(x), np.array(y), 10.0*np.log10(Sxx), cmap=cm.coolwarm)
plt.show()

类似的未回答问题：How to convert a spectrogram to 3d plot. Python

python 中的所需绘图，如 Matlab 的图形（此处最后一个绘图：https://www.mathworks.com/help/signal/ref/spectrogram.html）

Answer 1

你只需要让你的阵列形状正确：

fs = 11240.
t = 10
time = np.arange(fs*t) / fs
frequency = 1000.
mysignal = np.sin(2.0 * np.pi * frequency * time)

nperseg = 2**14
noverlap = 2**13
f, t, Sxx = signal.spectrogram(mysignal, fs, nperseg=nperseg,noverlap=noverlap)

myfilter = (f>800) & (f<1200)

f = f[myfilter]
Sxx = Sxx[myfilter, ...]

fig = plt.figure()
ax = fig.gca(projection='3d')

ax.plot_surface(f[:, None], t[None, :], 10.0*np.log10(Sxx), cmap=cm.coolwarm)
plt.show()

Answer 2

这是@Nils Werner 的回答的改编版本，其中包含不同的变量名和一组完整的导入。

import numpy as np
import matplotlib.pyplot as plt 
from scipy import signal # spectrogram function
from matplotlib import cm # colour map

# basic config
sample_rate = 11240.  # 
sig_len_secs = 10
frequency = 2000.

# generate the signal
timestamps_secs = np.arange(samplerate*sig_len_secs) / sample_rate
mysignal = np.sin(2.0 * np.pi * frequency * timestamps_secs) 

# extract the spectrum
freq_bins, timestamps, spec = signal.spectrogram(mysignal, sample_rate)

# 3d plot
fig = plt.figure()
ax = fig.gca(projection='3d')
ax.plot_surface(freq_bins[:, None], timestamps[None, :], 10.0*np.log10(spec), cmap=cm.coolwarm)
plt.show()

Python 3D 频谱图（类似于 matlab 的频谱图函数）

Python spectrogram in 3D (like matlab's spectrogram function)

python

matplotlib

scipy

fft

spectrogram