app.py

from TTS.api import TTS
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2", gpu=True).to("cuda")
# Dependencies
%cd /content/

import locale
locale.getpreferredencoding = lambda: "UTF-8"

!git clone https://github.com/justinjohn0306/Wav2Lip
!cd Wav2Lip && pip install -r requirements_colab.txt

%cd /content/Wav2Lip

!wget "https://www.adrianbulat.com/downloads/python-fan/s3fd-619a316812.pth" -O "face_detection/detection/sfd/s3fd.pth"
!wget 'https://github.com/justinjohn0306/Wav2Lip/releases/download/models/wav2lip.pth' -O 'checkpoints/wav2lip.pth'
!wget 'https://github.com/justinjohn0306/Wav2Lip/releases/download/models/wav2lip_gan.pth' -O 'checkpoints/wav2lip_gan.pth'
!wget 'https://github.com/justinjohn0306/Wav2Lip/releases/download/models/resnet50.pth' -O 'checkpoints/resnet50.pth'
!wget 'https://github.com/justinjohn0306/Wav2Lip/releases/download/models/mobilenet.pth' -O 'checkpoints/mobilenet.pth'
import subprocess
import assemblyai as aai
import requests
import moviepy.editor as mp

class translation:
  def __init__(self,video_path,original_language,target_language):
    self.video_path= video_path
    self.original_language = original_language
    self.target_language = target_language



  def org_language_parameters(self,original_language):
        if original_language == 'English':
            self.lan_code='en'
        elif original_language =='German':
            self.lan_code='de'
        elif original_language =='French':
            self.lan_code='fr'
        elif original_language =='Spanish':
            self.lan_code='es'
        else:
            self.lan_code = ''

  def target_language_parameters(self,target_language):
        if target_language == 'English':
            self.tran_code='en'
        elif target_language =='German':
            self.tran_code='de'
        elif target_language =='French':
            self.tran_code='fr'
        elif target_language =='Spanish':
            self.tran_code='es'
        else:
            self.tran_code = ''

  def extract_audio(self):
          video = mp.VideoFileClip(self.video_path)
          audio = video.audio
          audio_path = "output_audio.wav"
          audio.write_audiofile(audio_path)
          print("Audio extracted successfully!")
          return audio_path


  def transcribe_audio(self,audio_path):
      aai.settings.api_key = "c29eb650444a4ae4be6a787ebb15d5e2"
      config = aai.TranscriptionConfig(language_code=self.lan_code)
      transcriber = aai.Transcriber(config=config)
      transcript = transcriber.transcribe(audio_path)
      transcript_text = transcript.text
      return transcript_text
      if transcript.status == aai.TranscriptStatus.error:
          print(transcript.error)
          return None


  def translate_text(self,transcript_text):
          base_url = "https://api.cognitive.microsofttranslator.com"
          endpoint = "/translate"
          headers = {
              "Ocp-Apim-Subscription-Key": "cd226bb1f3644276bea01d82dd861cbb",
              "Content-Type": "application/json",
              "Ocp-Apim-Subscription-Region": "southeastasia"
          }
          params = {
              "api-version": "3.0",
              "from": self.lan_code,
              "to": self.tran_code
          }
          body = [{"text": transcript_text}]
          response = requests.post(base_url + endpoint, headers=headers, params=params, json=body)
          response.raise_for_status()
          translation = response.json()[0]["translations"][0]["text"]
          return translation




        #generate audio
  def generate_audio(self,translated_text):
      tts.tts_to_file(translated_text,
          speaker_wav='output_audio.wav',
          file_path="output_synth.wav",
          language= self.tran_code
      )
      return "output_synth.wav"

  def translate_video(self):
    audio_path = self.extract_audio()
    self.org_language_parameters(self.original_language)
    self.target_language_parameters(self.target_language)
    transcript_text = self.transcribe_audio(audio_path)
    translated_text = self.translate_text(transcript_text)
    translated_audio_path = self.generate_audio(translated_text)
      #Generate video
    %cd /content/Wav2Lip

      #This is the detection box padding, if you see it doesnt sit quite right, just adjust the values a bit. Usually the bottom one is the biggest issue
    pad_top =  0
    pad_bottom =  15
    pad_left =  0
    pad_right =  0
    rescaleFactor =  1
    video_path_fix = f"'../{self.video_path}'"
    audio_path_fix = f"'../{translated_audio_path}'"
    !python inference.py --checkpoint_path 'checkpoints/wav2lip_gan.pth' --face $video_path_fix --audio $audio_path_fix --pads $pad_top $pad_bottom $pad_left $pad_right --resize_factor $rescaleFactor --nosmooth --outfile '/content/output_video.mp4'
    return '/content/output_video.mp4'  # Return the path to the translated video file
    # return '/content/output_video.mp4', open('/content/output_video.mp4', 'rb')  # Return the path and file object of the translated video file

# from translator import translation  # Import the Translator class from translate module
import gradio as gr
import os
from google.colab import files


def app(video_path, original_language, target_language):
    %cd /content/

    video_name = os.path.basename(video_path)

    # Save the uploaded file to the content folder in Colab
    with open(video_name, "wb") as f:
        with open(video_path, "rb") as uploaded_file:
            f.write(uploaded_file.read())


    translator = translation(video_name, original_language, target_language)
    video_file = translator.translate_video()


    return video_file

interface_video_file = gr.Interface(
    fn=app,
    inputs=[
        gr.Video(label="Video Path"),
        gr.Dropdown(["English", "German", "French", "Spanish"], label="Original Language"),
        gr.Dropdown(["English", "German", "French", "Spanish"], label="Targeted Language"),
    ],
    outputs=gr.Video(label="Translated Video")
)

interface_video_file.launch(debug=True)