Notes

import json
from deepgram import Deepgram
from pydub import AudioSegment
import re
import wave
import numpy as np

DEEPGRAM_API_KEY = 'f0ca2acc45aca377f846be8eb2221c01945d1f4e'
PATH_TO_FILE = "/home/admin1/Desktop/test_audio/word1.wav"



PROFANE_WORDS = ["fucking", "bastard"] # Replace this with your list of profane words

# Load the beep sound
beep_sound_path = "/home/admin1/Downloads/censor-beep-01.wav"
beep_sound = AudioSegment.from_wav(beep_sound_path)

with wave.open(beep_sound_path, 'rb') as wav_file:
# Get the frame rate (frames per second) of the beep sound
frame_rate = wav_file.getframerate()
print("frame_rate:", frame_rate)


def censor_profane_words(transcript):
censored_audio = AudioSegment.empty()
cursor = 0 # Track the cursor position in the audio

for profane_word in PROFANE_WORDS:
word_occurrences = [m.start() for m in re.finditer(re.escape(profane_word), transcript, re.IGNORECASE)]
print("word_occurrences", word_occurrences)
i = 0
for start_time in word_occurrences:
print("Start time:", start_time)
# Calculate the duration of the profane word in milliseconds
duration_ms = len(profane_word) * (1000 / frame_rate)
print("duration_ms:", duration_ms)
print("word_length:", len(profane_word))
# Add non-profane audio before the profane word
non_profane_audio = AudioSegment.from_wav(PATH_TO_FILE)[cursor:int(start_time * frame_rate / 1000)]
# print("cursor:",cursor)
# print(int(start_time * frame_rate / 1000))
# print("non_profane_audio:", non_profane_audio)

censored_audio += non_profane_audio
print("censored_audio1:", censored_audio)

# Add beep sound for the profane word duration
censored_audio += beep_sound
print("censored_audio12:",censored_audio)
# print("beep_sound:",beep_sound)
# Update the cursor position
cursor = int(start_time * frame_rate / 1000) + int(2*duration_ms)
print("z:",int(start_time * frame_rate / 1000))
print("z1:",int(2*duration_ms))

print("cursor:", cursor)
i=i+1
# Add any remaining non-profane audio after the last profane word
remaining_audio = AudioSegment.from_wav(PATH_TO_FILE)[cursor:]
print("remaing_audio:", remaining_audio)
censored_audio += remaining_audio
print("censored_audio:", censored_audio)

return censored_audio


def main():
deepgram = Deepgram(DEEPGRAM_API_KEY)
with open(PATH_TO_FILE, 'rb') as audio_file:
source = {'buffer': audio_file, 'mimetype': 'audio/wav'}
response = deepgram.transcription.sync_prerecorded(source, {'punctuate': True})



if 'results' in response and 'channels' in response['results']:
transcript = response['results']['channels'][0]['alternatives'][0]['transcript']
censored_audio = censor_profane_words(transcript)
censored_audio.export("/home/admin1/Downloads/censored_output.wav", format="wav")
print("Censored audio saved successfully.")
else:
print("Transcription not available in the response.")


if __name__ == "__main__":
main()