Notes

# -*- coding: utf-8 -*-
"""
Created on Sat Jan 18 20:00:49 2020

@author: Gulshan Rana
"""

from google.cloud import speech_v1
from google.cloud.speech_v1 import enums
import io
import os
#import audiolab
import numpy as np
from scipy.io.wavfile import write
#from subprocess import call
from pydub import AudioSegment

def sample_recognize(local_file_path):

client = speech_v1.SpeechClient.from_service_account_json('PrestiSolutions-fb3de2002b43.json')

language_code = "en-IN"
sample_rate_hertz = 48000
enable_word_time_offsets = True
profanity_filter= True

encoding = enums.RecognitionConfig.AudioEncoding.FLAC

config = {
"enable_word_time_offsets": enable_word_time_offsets,
"language_code": language_code,
"sample_rate_hertz": sample_rate_hertz,
"encoding": encoding,
"profanity_filter": profanity_filter
}
#reading the audio file
with io.open(local_file_path, "rb") as f:
content = f.read()
audio = {"content": content} # audio ready to be propcessed
#recognising the profaned transcript
response = client.recognize(config, audio)


start_time=[]
end_time=[]
timeline=[]
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
print(u"Transcript: {}".format(alternative.transcript))
for word in alternative.words:
timeline.append(word.start_time.seconds+word.start_time.nanos*(10**-9))
timeline.append(word.end_time.seconds+word.end_time.nanos*(10**-9))
if '*' in word.word:
start_time.append(word.start_time.seconds+word.start_time.nanos*(10**-9))
end_time.append(word.end_time.seconds+word.end_time.nanos*(10**-9))
return start_time,end_time,timeline

input_audio='inaudio.aac'

os.system(('ffmpeg -i {} -vn out1.flac mono.wav ').format(input_audio)) # for converting format
os.system('ffmpeg -i out1.flac -ac 1 mono.flac ') #for creating single channel of audio
# ffmpeg -i xyz1.wav -ss 0 -t 3 -acodec copy sound0.wav
#audio_path="mono.wav"
print(1)
start_time,end_time,timeline= sample_recognize('mono.flac')
print(2)
#scr_tst.py()




#Generating beep at profanity
sps = 44100
freq_hz = 1000.0
duration = 0.3
vol = 0.3
esm = np.arange(duration * sps)
wf = np.sin(2 * np.pi * esm * freq_hz / sps)
wf_quiet = wf * vol
wf_int = np.int16(wf_quiet * 32767)
write("beep.wav", sps, wf_int)


start_time,end_time
init_=0.0
len_=len(start_time)
def split_profaned_audio(audio_path,start_time,end_time,ti_me):
audio_path==audio_path
start_time=start_time
end_time=end_time
audio_file="audio"+str(ti_me+1)+".wav"
audio_file = r"%s"%audio_file
os.system(('ffmpeg -i {} -ss {} -t {} -acodec copy {}').format(audio_path,start_time,end_time-start_time,audio_file))
print('done')

audio_path='mono.wav'

print('split0 ok')

split_profaned_audio(audio_path,timeline[0]+0.1,start_time[0]+0.1,-1) #initialisation
#call(["ffmpeg","-i", "xyz1.wav", "-ss", "0", "-t", "start_time[1]", "-acodec", "copy", "sound0.wav"])
#sound0 = open("audio0.flac", "rb").read()
#beep1 = open("beep.flac", "rb").read()
#combined_sounds = sound0 + beep1
#combined_sounds = open("combined_sounds.flac", "wb").write(combined_sounds)
sound = AudioSegment.from_wav("audio0.wav")
beep = AudioSegment.from_wav("beep.wav")
combined_sounds = sound + beep
combined_sounds.export("combined_sounds.wav", format="wav")


if len_!=1:
for ti_me in range(len_-1):
split_profaned_audio(audio_path,end_time[ti_me]+0.1,start_time[ti_me+1]+0.1,ti_me) #initialisation
#call(["ffmpeg","-i", "xyz1.wav", "-ss", "0", "-t", "start_time[1]", "-acodec", "copy", "sound0.wav"])
#sound="audio"+str(ti_me+1)+".flac"
#sound = r"%s"%sound
#sound = open(sound, "rb").read()

#sound = open("sound.flac", "wb").write(sound)
#combined_sounds = combined_sounds+sound + beep1
#combined_sounds = open("combined_sounds.flac", "wb").write(combined_sounds)
#combined_sounds= open("combined_sounds.flac", "rb").read()
for ti_me in range(len_-1):
'''beep1 = open("beep.flac", "rb").read()
sound="audio"+str(ti_me+1)+".flac"
sound = r"%s"%sound
sound = open(sound, "rb").read()
sound1a=sound+beep1
sound1 = open("sound1a.flac", "wb").write(sound1a)

sound1 = open("sound1.flac", "rb").read()
c1 = open(combined_sounds, "rb").read()
combined_sounds=c1+sound1


sound="audio"+str(ti_me+1)+".flac"
input_audio = r"%s"%sound
sound="audio"+str(ti_me+1)+".wav"
output_audio = r"%s"%sound
os.system(('ffmpeg -i {} -vn {} ').format(input_audio,output_audio))
'''
sound="audio"+str(ti_me+1)+".wav"
input_audio = r"%s"%sound
beep = AudioSegment.from_wav("beep.wav")
sub_sound = AudioSegment.from_wav(input_audio)

combined_sounds=combined_sounds+sub_sound+beep

combined_sounds.export("combined_sounds.wav", format="wav")

print('winner')

else:
ti_me=0
split_profaned_audio(audio_path,end_time[ti_me]+0.1,timeline[-1]+0.1,ti_me) #initialisation
'''call(["ffmpeg","-i", "xyz1.wav", "-ss", "0", "-t", "start_time[1]", "-acodec", "copy", "sound0.wav"])
sound="audio"+str(ti_me+1)+".flac"
sound = r"%s"%sound
sound = open(sound, "rb").read()
sound = open("sound.flac", "wb").write(sound)
combined_sounds = combined_sounds+sound + beep1
combined_sounds = open("combined_sounds.flac", "wb").write(combined_sounds)
combined_sounds= open("combined_sounds.flac", "rb").read()'''
sound="audio"+str(ti_me+1)+".flac"
input_audio = r"%s"%sound
sub_sound = AudioSegment.from_wav("input_audio.wav")

combined_sounds=combined_sounds+sub_sound+beep

combined_sounds.export("combined_sounds.wav", format="wav")
print('else_winner')


split_profaned_audio(audio_path,end_time[-1]+0.1,timeline[-1]+0.1,len_-1)
sound="audio"+str(len_)+".wav"
input_audio = r"%s"%sound
sub_sound = AudioSegment.from_wav(input_audio)


combined_sounds=combined_sounds+sub_sound

combined_sounds.export("combined_sounds.wav", format="wav")
'''
split_profaned_audio(audio_path,7.5,2.5,0)
sound1 = AudioSegment.from_wav("audio1.wav")
combined_sounds = combined_sounds + sound1 + beep
combined_sounds.export("finaudF0.0.wav", format="wav")
print('aok')



for ti_me in range(len_-1):
split_profaned_audio(audio_path,start_time[ti_me+1],end_time[ti_me],ti_me+1)
audio_file="audio"+str(ti_me+1)+".wav"
audio_file = r"%s"%audio_file
audio = AudioSegment.from_wav(audio_file)
audio=audio+beep
combined_sounds = combined_sounds +audio

combined_sounds = combined_sounds + beep
length_of_audio=timeline[-1]
split_profaned_audio(audio_path,end_time[-1],length_of_audio,len_)
audio1=AudioSegment.from_wav("{}").format("audio"+str(len_+1)+".wav")
combined_sounds=combined_sounds+audio1
combined_sounds.export("finaudF.0.wav", format="wav")

print(3)


'''