em474re/CovidSpeechChallenge_2021
2
1
Fork 0
Code Issues Pull Requests Releases Wiki Activity

handcrafted features

Browse Source
This commit is contained in:
em474re 2021-09-07 09:48:00 +02:00
parent 6b3fbce6d1
commit 71e262b340
1 changed files with 14 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

31
extract_handcrafted_features.py → src/extract_handcrafted_features.py
View File

@ -1,5 +1,4 @@
import json import os
import os
import sys import sys
import warnings import warnings
from math import pi from math import pi
@ -13,7 +12,7 @@ warnings.filterwarnings("ignore")
SR = 22050 # sample rate SR = 22050 # sample rate
FRAME_LEN = int(SR / 10) # 100 ms FRAME_LEN = int(SR / 10) # 100 ms
HOP = int(FRAME_LEN / 2) # 50% overlap, meaning 5ms hop length HOP = int(FRAME_LEN / 2) # 50% overlap, meaning 5 ms hop length (step size)
MFCC_dim = 13 # the MFCC dimension MFCC_dim = 13 # the MFCC dimension
def sta_fun(np_data): def sta_fun(np_data):
@ -190,16 +189,17 @@ def extract_features(signal, signal_sr):
if __name__ == "__main__": if __name__ == "__main__":
# data path (raw_files\devel OR test OR train folder) # data path (dist/wav/)
path = sys.argv[1] path = sys.argv[1]
# train, devel or test subset
dataset = sys.argv[2]
x_data = [] x_data = []
y_label = [] y_label = []
y_uid = []
#extract features # extract features for all audio samples from correct subset
files = os.listdir(path) for file in sorted([f for f in os.listdir(path) if dataset in f]):
for file in files:
try: try:
sample_path = os.path.join(path,file) sample_path = os.path.join(path,file)
file_b = sample_path file_b = sample_path
@ -213,18 +213,11 @@ if __name__ == "__main__":
yt, index = librosa.effects.trim( yt, index = librosa.effects.trim(
y, frame_length=FRAME_LEN, hop_length=HOP y, frame_length=FRAME_LEN, hop_length=HOP
) )
duration = librosa.get_duration(y=yt, sr=sr)
if duration < 2:
continue
features = extract_features(signal=y, signal_sr=sr) features = extract_features(signal=y, signal_sr=sr)
x_data.append(features.tolist()) x_data.append(features.tolist())
#save features in numpy.array # save features and labels
x_data = np.array(x_data) df = pd.read_csv('./dist/lab/' + dataset + '.csv', sep =',')
labels_path = 'labels\\' + os.path.basename(os.path.normpath(path)) + '.csv' np.save(os.path.join('./features/hand_features',"x_" + dataset + "_data.npy"), np.array(x_data))
df = pd.read_csv(labels_path, sep =',') np.save(os.path.join('./features/hand_features',"y_" + dataset + "_label.npy"), df.label)
y_label = df.label
np.save(os.path.join('hand_features',"x_" + os.path.basename(os.path.normpath(path)) + "_data.npy"), x_data)
np.save(os.path.join('hand_features',"y_" + os.path.basename(os.path.normpath(path)) + "_label.npy"), y_label)