-
Notifications
You must be signed in to change notification settings - Fork 0
/
audio2spectrogram.py
61 lines (48 loc) · 1.97 KB
/
audio2spectrogram.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
'''
This script is used to convert raw audio data into mel-spectrograms for image processing and
computer vision analysis of audio files. The script is based on EmoDB dataset, which can be found
at: https://www.kaggle.com/piyushagni5/berlin-database-of-emotional-speech-emodb
'''
import os
import shutil
import numpy as np
import matplotlib.pyplot as plt
import librosa
import librosa.display
from scipy.io import wavfile
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('--audio_dir',type=str, help='path to directory containing audio files per emotion class')
parser.add_argument('--data_dir', type=str, default=os.getcwd(), help='path where images will be stored')
args = parser.parse_args()
AUDIO_DIR = args.audio_dir
IMG_DIR = args.data_dir
# save audios as spectrograms
for AUDIO_FILE in os.listdir(AUDIO_DIR):
samples, sample_rate = librosa.load(os.path.join(AUDIO_DIR, AUDIO_FILE), sr=None)
sgram = librosa.stft(samples)
sgram_mag, _ = librosa.magphase(sgram)
mel_scale_sgram = librosa.feature.melspectrogram(S=sgram_mag, sr=sample_rate)
mel_sgram = librosa.amplitude_to_db(mel_scale_sgram, ref=np.min)
librosa.display.specshow(mel_sgram, sr=sample_rate, x_axis='time', y_axis='mel')
plt.colorbar(format='%+2.0f dB').remove()
plt.axis('off')
plt.savefig(f"./{IMG_DIR}/{AUDIO_FILE}.png")
### REPLACE THE LABEL:EMOTION DICTIONARY WITH THAT RELEVANT TO DATASET ###
label2name = {
"L": "Boredom",
"A": "Fear",
"E": "Disgust",
"F": "Happiness",
"T": "Sadness",
"W": "Anger",
"N": "Neutral"
}
# move the spectrograms to respective class folders
for cat in label2name.keys():
os.mkdir(os.path.join(IMG_DIR, label2name[cat]))
for filename in os.listdir(IMG_DIR):
if os.path.isfile(IMG_DIR+"/"+filename) and filename[:-3] == "png":
cat = filename[5]
shutil.move(os.path.join(IMG_DIR, filename), os.path.join(IMG_DIR, label2name[cat]))
print("====> All audios converted to mel-spectrograms !!")