execution code for vocalsoundfilter.py

filter_vocal_sounds.py

#!python
# -*- coding: utf-8 -*-
import numpy as np
from random import shuffle, seed, randrange, choice, uniform
from math import floor

lines = []
data = []

low = 0x3000
high = 0x30FA
simple_cjk = high-low+1 # kana or punctuation
classes = simple_cjk+2 # one is below, one is above

with open("test.txt", "r", encoding="utf-8") as f:
    for line in f:
        lines += [line]
        real_line = []
        for c in line:
            char = [0]*classes
            num = ord(c)
            num -= low
            num += 1
            if num < 0:
                num = 0
            if num > classes-1:
                num = classes-1
            char[num] = 1
            real_line += [char]
        data += [np.array([real_line])]



from keras.models import load_model

model = load_model("model.h5")

positive = False

# necessary to prevent transcoding ～ and 〜
import sys
def print(string):
    sys.stdout.buffer.write(str(string).encode('utf-8'))
    sys.stdout.buffer.write("\n".encode('utf-8'))

for i in range(len(lines)):
    line = lines[i]
    if len(line) == 0:
        if positive:
            pass
        else:
            print(line)
    else:
        mydata = data[i]
        test = model.predict(mydata)[0]
        if (test[1] > test[0]) == positive:
            print(line.rstrip("\n"))

exit(0)