A Practical Guide to Spectrogram Analysis for Audio Signal Processing

TL;DR

This paper provides a practical guide to spectrogram analysis for audio signal processing, using finger-snapping as a case study to illustrate time-frequency analysis. It explains how to compute time-varying frequency content via spectrograms and PSD estimation with Welch's method, employing FFT with a Hanning window. A key contribution is demonstrating the impact of FFT segment size on PSD and spectrogram resolution across different sampling rates (44100 Hz and 96000 Hz) and providing concrete parameter recommendations. The work offers actionable guidance for practitioners on parameter choices to balance resolution and computational cost in real-world audio analysis.

Abstract

The paper summarizes spectrogram and gives practical application of spectrogram in signal processing. For analysis, finger-snapping is recorded with a sampling rate of 441000 Hz and 96000 Hz. The effects of the number of segments on the Power Spectral Density (PSD) and spectrogram are analyzed and visualized.

Figures (18)

• Figure 1: Fourier transform of the cosine function.
• Figure 2: Square periodic and sinusoid signals.
• Figure 3: Different harmonic sinusoids.
• Figure 4: Sinusoid approximation for square periodic wave.
• Figure 5: Frequency increase in time.