fft-graph.py

# plotting an audio file in both time-domain and frequency-domain
# oliver thurley, november 2023

# requires scipy, numpy, matplotlib libraries
# python v.3

import matplotlib.pyplot as plt
from scipy.io import wavfile
import numpy as np

###
### SET YOUR AUDIO FILE PATH HERE:
filename = "/path/to/audio.wav" #change this to your audio file

freq_limit = 8000 # Set clip upper limit (Hz) of the plots
############

def fftChart(filename):
    rate, wave = wavfile.read(filename) # only works with 16-bit WAV?
    # wave = wave[:, 0] #only the data from channel 0, we only want one channel # uncomment this line to convert file to mono

    duration = round(len(wave) / rate, 2) # convert samples to duration in secs
    t = np.arange(0.0, duration, 1/rate)
    
    print("filename: "+ str(filename))
    print("duration: "+ str(duration) + " seconds")

    ## PLOT CHARTS
    fig, (ax, magspc, spect) = plt.subplots(nrows=3, ncols=1, sharex=False, figsize=(12, 8))

    ## Time-domain plot
    timeSeq = np.linspace(0, duration, len(wave))
    ax.plot(timeSeq, wave, color="m", linewidth=1.5)
    ax.set(title=filename)
    ax.set_xlabel('time (s)',fontsize=8)
    ax.set_ylabel('amplitude',fontsize=8)
    ax.grid(linewidth=0.2)
    ax.set_yticks([]) # removes ticks
    ax.tick_params(left=False, bottom=False, labelbottom=True, labelsize='smaller')

    ## Do spectrum plot (magnitude spectrum)
    magspc.magnitude_spectrum(wave, Fs=rate, color='m')
    magspc.set_xlabel('frequency (Hz)',fontsize=8)
    magspc.set_ylabel('magnitude',fontsize=8)
    magspc.tick_params(left=False, bottom=False)
    magspc.set_yscale('linear')
    magspc.grid(linewidth=0.2)
    magspc.set_yticks([]) # removes ticks
    magspc.tick_params(left=False, bottom=False, labelsize='smaller')
    magspc.set_xlim([0, freq_limit])
    magspc.grid(linewidth=0.2)

    ## Spectrogram plot
    spect.specgram(wave, Fs=rate, NFFT=2048, cmap='magma') # NFFT= length of windowing segments
    spect.set_xlabel('time (s)',fontsize=8)
    spect.set_ylabel('frequency (Hz)',fontsize=8)
    spect.set_ylim([0, freq_limit])
    spect.tick_params(left=False, bottom=False, labelbottom=True, labelsize='smaller')

    plt.subplots_adjust(hspace=0.3)
    plt.savefig(filename+"-plot.png", dpi=300)
    # plt.show()

## Uncomment this section for batch processing audio files in a folder:
# import os, glob
# path = 'folder/path/'
# for filename in glob.glob(os.path.join(path, '*.wav')):
#     fftChart(filename)

## Run process on (single) filename:
fftChart(filename)