translator.py

import subprocess
import assemblyai as aai
import requests
import moviepy.editor as mp
from TTS.api import TTS

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 == 'Italian':
            self.lan_code = 'it'
        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 == 'Italian':
            self.tran_code = 'it'
        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)
        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)
        return transcript.text

    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)
        return response.json()[0]["translations"][0]["text"]

    def generate_audio(self, translated_text):
        tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2")
        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)

        subprocess.run(['python', 'inference.py', '--checkpoint_path', 'checkpoints/wav2lip_gan.pth', '--face', self.video_path, '--audio', translated_audio_path, '--outfile', 'output_video.mp4'])
        return 'output_video.mp4'