-
Notifications
You must be signed in to change notification settings - Fork 7
/
label_dataset.py
81 lines (73 loc) · 2.34 KB
/
label_dataset.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
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
'''
The purpose of this script is to create a dataset for training the
emotion analysis model. It does the following:
1. Reads all the audio files placed in a folder, breaks it into chunks of 30 seconds each.
2. Plays each chunk and inputs user input for the emotion.
3. Saves the audio chunk as a .wav file with its filename in this format - "<Emotion>_<count>.wav"
'''
from __future__ import division
import pyaudio
import scipy.io.wavfile as wav
import math
from os import listdir
from os.path import isfile, join
segments = [] # A list of 30 sec segments.
path = "calls" # The location where all the audio files are initially stored.
c_total = 0
c_read = 0
c_skipped = 0
'''
The following part reads each audio file, breaks it into 30 seconds segments and appends
each segment into a list.
'''
audio_files = [f for f in listdir(path) if isfile(join(path, f))]
for audio_file in audio_files:
c_total+=1
print "Reading file: "+str(c_total)
try:
rate_sig, sig = wav.read(join(path, audio_file))
c_read+=1
except:
c_skipped+=1
continue
for i in range(int(math.ceil(len(sig)/(rate_sig*30)))):
segments.append(sig[i*(rate_sig*30):i*(rate_sig*30) + rate_sig*30])
print "-"*50
print "\n\n"
print "Total files: "+str(c_total)
print "Read files: "+str(c_read)
print "Skipped files:"+str(c_skipped)
print "Total 30 sec chunks: "+str(len(segments))
print "-"*50
'''
The following part plays each segment and accepts user input for each segment as follows:
1 <-> 'negative'
2 <-> 'neutral'
3 <-> 'positive'
'''
FORMAT = pyaudio.paInt16
CHANNELS = 1
CHUNK = 1024
map_key_to_emotion = {'1':'negative','2':'neutral', '3':'positive'}
p = pyaudio.PyAudio()
#opens an audio stream
stream = p.open(format = FORMAT,
channels = CHANNELS,
rate = rate_sig,
output = True, frames_per_buffer=rate_sig)
#writes into the stream
count_label = 519
for segment in segments[520:]:
count_label+=1
stream.write(segment[0:int(len(segment)/2)])
stream.write(segment[int(len(segment)/2):])
try:
segment_label = map_key_to_emotion[raw_input(str(count_label)+"> ")]
except:
continue
wav.write("data\\"+segment_label+"_"+str(count_label)+".wav", rate_sig, segment)
#stops the stream
stream.stop_stream()
stream.close()
#close PyAudio
p.terminate()