Trying to create Array Channels

[glaure]

Hi,

I am trying the create "Array Channels" like described here:

https://www.asam.net/standards/detail/mdf/wiki/

[CG]
    - [CN]
      [CN]
       - [CA]

The given mf4_matrix_demo creates two channels.
One holds the result of an FFT, the other a spectrogram. The second one is also visualized with mathplotlib.

This is the way I tried:

import numpy as np
from asammdf import MDF, Signal
from scipy.signal import spectrogram
from scipy.io import wavfile
import matplotlib.pyplot as plt
import os
import sys

cycles = 100
sigs = []

mdf = MDF()

t = np.arange(cycles, dtype=np.float64)

# FFT spectrum for intensity diagram
fft_spectrum = np.abs(np.fft.fft2(np.sin(2 * np.pi * t / cycles).reshape((10, 10))))
sig = Signal(
    fft_spectrum,
    t[:10],  # Assuming the time base for the intensity diagram is the first 10 time points
    name="Channel_FFT_Spectrum",
    unit="Intensity",
    comment="FFT spectrum for intensity diagram",
)
sigs.append(sig)

# Spectrogram for order analysis
sf, audio = wavfile.read(os.path.join(os.path.dirname(sys.argv[0]), 'example.wav'))
if audio.ndim > 1:
    audio = np.mean(audio, axis=1)  # Ensure audio is a 1D array if it is multi-channel
frequencies, times, Sxx = spectrogram(audio, sf, scaling='density')
plt.pcolormesh(times, frequencies, np.log10(Sxx))
plt.ylabel('Frequency [Hz]')
plt.xlabel('Time [sec]')
plt.show()

# No need to transpose Sxx as we want frequencies on the x-axis
sig = Signal(
    Sxx,
    frequencies,
    name="Channel_Spectrogram",
    unit="dB",
    comment="Spectrogram for order analysis",
)
sigs.append(sig)

mdf.append(sigs, comment="matrix")

mdf.save("demo_matrix.mf4", overwrite=True)

MDF Validator displays it like this. I do not see the array information I would like to have. No dimensions etc

What am I missing?

My motivation is to export spectral data to FlexPro and currently failing at that.
FlexPro only seems to see 1 vector sample instead of the whole spectrum.

Best regards
Gunther

Update:
The script needs a example.wav file

example.zip

[danielhrisca]

It is not possible to have the master channel as frequency\Hz, it will show up as time|seconds

import numpy as np
from asammdf import MDF, Signal
from scipy.signal import spectrogram
from scipy.io import wavfile
import matplotlib.pyplot as plt
import os
import sys

cycles = 100
sigs = []

mdf = MDF()

t = np.arange(cycles, dtype=np.float64)

# FFT spectrum for intensity diagram
fft_spectrum = np.abs(np.fft.fft2(np.sin(2 * np.pi * t / cycles).reshape((10, 10))))

types = [
    ("Channel_FFT_Spectrum", "<f8", fft_spectrum.shape[1:]),  # same name as the signal name
]
fft_spectrum = np.rec.fromarrays([fft_spectrum], dtype=np.dtype(types))

sig = Signal(
    fft_spectrum,
    t[:10],  # Assuming the time base for the intensity diagram is the first 10 time points
    name="Channel_FFT_Spectrum",
    unit="Intensity",
    comment="FFT spectrum for intensity diagram",
)
# append the signals separately
mdf.append(sig, comment="FFT Spectrum")

# Spectrogram for order analysis
sf, audio = wavfile.read(os.path.join(os.path.dirname(sys.argv[0]), r'd:\TMP\output.wav'))
if audio.ndim > 1:
    audio = np.mean(audio, axis=1)  # Ensure audio is a 1D array if it is multi-channel
frequencies, times, Sxx = spectrogram(audio, sf, scaling='density')
plt.pcolormesh(times, frequencies, np.log10(Sxx))
plt.ylabel('Frequency [Hz]')
plt.xlabel('Time [sec]')
plt.show()

types = [
    ("Channel_Spectrogram", "<f4", fft_spectrum.shape[1:]),  # same name as the signal name
]
Sxx = np.rec.fromarrays([Sxx], dtype=np.dtype(types))

# No need to transpose Sxx as we want frequencies on the x-axis
sig = Signal(
    Sxx,
    frequencies,
    name="Channel_Spectrogram",
    unit="dB",
    comment="Spectrogram for order analysis",
)

mdf.append(sig, comment="Channel spectrogram")

mdf.save("demo_matrix.mf4", overwrite=True)

[glaure]

Hi,

thank you for your fast response.
Indeed your changes do now create Channel Arrays.
Still I see that the dimension of the spectrogram is 1 x 1169 instead of the 129x1169 matrix I expected:
ca_ndim = 1
ca_dim_size[0] = 1169

Do you have an idea why?

bye Gunther