Spectrogram generation in SciPy - Mini Project: Build & Apply

Spectrogram Generation with SciPy

📖 Scenario: Imagine you are analyzing sound recordings to understand their frequency content over time. A spectrogram is a useful tool that shows how the frequencies in a sound change as time passes.In this project, you will use Python's scipy library to generate a spectrogram from a simple sound signal.

🎯 Goal: You will create a simple sound signal, configure parameters for the spectrogram, generate the spectrogram using SciPy, and finally display the spectrogram data.

📋 What You'll Learn

Create a numpy array representing a sound signal

Set a sampling frequency variable

Use SciPy's spectrogram function to compute the spectrogram

Print the shape of the spectrogram data

💡 Why This Matters

🌍 Real World

Spectrograms are widely used in audio processing, speech analysis, and music technology to study how sound frequencies vary over time.

💼 Career

Understanding spectrogram generation is useful for roles in audio engineering, data analysis, and machine learning involving time-series or audio data.

1

Create a simple sound signal

Import numpy as np. Create a variable called fs and set it to 1000 to represent the sampling frequency in Hertz. Create a time array called t using np.linspace from 0 to 1 second with fs points. Create a signal called signal as a sine wave with frequency 5 Hz using np.sin(2 * np.pi * 5 * t).

SciPy

# Import numpy as np
# Create fs = 1000
# Create t using np.linspace from 0 to 1 with fs points
# Create signal as sine wave with frequency 5 Hz

Need a hint?

Use np.linspace to create the time array and np.sin to create the sine wave.

2

Import spectrogram function and set parameters

Import spectrogram from scipy.signal. Create a variable called nperseg and set it to 100 to define the length of each segment for the spectrogram.

SciPy

import numpy as np
fs = 1000
t = np.linspace(0, 1, fs, endpoint=False)
signal = np.sin(2 * np.pi * 5 * t)
# Import spectrogram from scipy.signal
# Set nperseg = 100

Need a hint?

Use from scipy.signal import spectrogram to import the function.

3

Generate the spectrogram

Use the spectrogram function with arguments signal, fs, and nperseg. Assign the returned values to variables frequencies, times, and Sxx.

SciPy

import numpy as np
fs = 1000
t = np.linspace(0, 1, fs, endpoint=False)
signal = np.sin(2 * np.pi * 5 * t)
from scipy.signal import spectrogram
nperseg = 100
# Generate spectrogram and assign to frequencies, times, Sxx

Need a hint?

Call spectrogram(signal, fs, nperseg=nperseg) and unpack the result.

4

Display the spectrogram shape

Print the shape of the spectrogram data array Sxx using print(Sxx.shape).

SciPy

import numpy as np
fs = 1000
t = np.linspace(0, 1, fs, endpoint=False)
signal = np.sin(2 * np.pi * 5 * t)
from scipy.signal import spectrogram
nperseg = 100
frequencies, times, Sxx = spectrogram(signal, fs, nperseg=nperseg)
# Print the shape of Sxx

Need a hint?

Use print(Sxx.shape) to see the size of the spectrogram matrix.