translate.py

import os
import moviepy.editor as mp
import assemblyai as aai
import requests
import azure.cognitiveservices.speech as speechsdk
from moviepy.editor import AudioFileClip
from gradio_client import Client

class Translate:
    def __init__(self, video_path, target_language,original_language,speaking_rate):
        self.video_path = video_path
        self.target_language = target_language
        self.original_language=original_language
        self.aai_api_key = "c29eb650444a4ae4be6a787ebb15d5e2"
        self.txtospech_key = "358c77527e48454cbf5bf2bd54f03161"
        self.translation_api_key = "394833878dd54214886cd81a35ac35dc"
        self.spechtotxt_key = "07ac642da789462d87ad47a790ec6d5f"
        self.speaking_rate= speaking_rate

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

    def gender_detection(self):
        gender_model_url = "https://salman11223-gender-detection.hf.space/--replicas/wml9f/"
        gender_client = Client(gender_model_url)
        gender = gender_client.predict(
            'audio.wav', api_name="/predict"
        )
        print(gender)
        return gender



    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 set_language_parameters(self, target_language, detected_gender):
            if target_language == 'English':
                self.language_code = 'en-US'
                self.trans_code = 'en'
                self.voice_names = 'en-US-GuyNeural' if detected_gender == 'male' else 'en-US-AriaNeural'
            elif target_language == 'German':
                self.language_code = 'de-DE'
                self.trans_code = 'de'
                self.voice_names = 'de-DE-ConradNeural' if detected_gender == 'male' else 'de-DE-KatjaNeural'
            elif target_language == 'French':
                self.language_code = 'fr-CA'
                self.trans_code = 'fr'
                self.voice_names = 'fr-CA-JeanNeural' if detected_gender == 'male' else 'fr-CA-SylvieNeural'
            elif target_language == 'Spanish':
                self.language_code = 'es-ES'
                self.trans_code = 'es'
                self.voice_names = 'es-ES-AlvaroNeural' if detected_gender == 'male' else 'es-ES-ElviraNeural'
            elif target_language == 'Urdu':
                self.language_code = 'ur-PK'
                self.trans_code = 'ur'
                self.voice_names = 'ur-PK-AsadNeural' if detected_gender == 'male' else 'ur-PK-UzmaNeural'
            else:
                # Handle unsupported languages or set default values
                self.voice_names = []
                self.language_code = ''
                self.trans_code = ''


            print("Target Language:", target_language)
            print("Trans Code:", self.trans_code)

    def get_voice_names(self):
        return self.voice_names

    def get_language_code(self):
        return self.language_code



    def get_audio_duration(self, audio_path):
        audio_clip = AudioFileClip(audio_path)
        audio_duration = audio_clip.duration
        return audio_duration

    def transcribe_audio(self, audio_path):
        aai.settings.api_key = self.aai_api_key
        config = aai.TranscriptionConfig(self.lan_code)
        transcriber = aai.Transcriber(config=config)
        transcript = transcriber.transcribe(audio_path)
        file_path = "transcript.srt"
        filepath = "t.txt"
        with open(file_path, "w") as file:
            file.write(transcript.export_subtitles_srt())
        with open(filepath, "w") as file:
            file.write(transcript.text)


    def generate_ssml(self, text, speaking_rate):
        # Construct SSML with the given text, speaking rate, voice name, and language code
        return f'<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="{self.language_code}"><voice name="{self.voice_names}"><prosody rate="{speaking_rate}">{text}</prosody></voice></speak>'
    def text_to_speech(self, text, apikey, reggion, out_aud_file, speaking_rate):
        ssml = self.generate_ssml(text, speaking_rate)
        speech_config = speechsdk.SpeechConfig(subscription=apikey, region=reggion)
        audio_config = speechsdk.audio.AudioOutputConfig(filename=out_aud_file)
        speech_synthesizer = speechsdk.SpeechSynthesizer(speech_config=speech_config, audio_config=audio_config)
        speech_synthesizer.speak_ssml_async(ssml).get()

    def translate_text(self, text):
        base_url = "https://api.cognitive.microsofttranslator.com"
        endpoint = "/translate"
        headers = {
            "Ocp-Apim-Subscription-Key": self.translation_api_key,
            "Content-Type": "application/json",
            "Ocp-Apim-Subscription-Region": "southeastasia"
        }

        params = {
            "api-version": "3.0",
            "to": self.trans_code
        }
        body = [{"text": 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

    def transcribe_and_translate(self):
        audio_path = self.extract_audio()
        self.org_language_parameters(self.original_language)
        self.transcribe_audio(audio_path)
        gender = self.gender_detection()
        print("Detected Gender:", gender)
        self.set_language_parameters(self.target_language,gender)
        with open("transcript.srt", 'r') as srt_file:
            original_srt_content = srt_file.read()

        original_subtitles = original_srt_content.strip().split('\n\n')

        translated_subtitles = []
        for subtitle in original_subtitles:
            lines = subtitle.split('\n')
            sequence_number = lines[0]
            timestamp = lines[1]
            original_text = '\n'.join(lines[2:])
            translated_text = self.translate_text(original_text)
            translated_subtitle = f"{sequence_number}\n{timestamp}\n{translated_text}"
            translated_subtitles.append(translated_subtitle)

        translated_srt_content = '\n\n'.join(translated_subtitles)
        translated_srt_path = "translated_file.srt"

        with open(translated_srt_path, 'w', encoding='utf-8') as srt_file:
            srt_file.write(translated_srt_content)

        # Loop through each translated subtitle and generate speech
        translated_audio_paths = []
        for subtitle in translated_subtitles:
            lines = subtitle.split('\n')
            sequence_number = lines[0]
            timestamp = lines[1]
            translated_text = '\n'.join(lines[2:])
            translated_audio_path = f"translated_audio_{sequence_number}.wav"
            self.text_to_speech(translated_text, self.txtospech_key, "southeastasia", translated_audio_path, self.speaking_rate)
            translated_audio_paths.append(translated_audio_path)

        # Create a list to store the audio clips
        translated_audio_clips = []

        # Loop through each translated audio path and create an AudioFileClip
        for audio_path in translated_audio_paths:
            translated_audio_clip = mp.AudioFileClip(audio_path)
            translated_audio_clips.append(translated_audio_clip)

        # Concatenate the translated audio clips into a single audio file
        translated_audio = mp.concatenate_audioclips(translated_audio_clips)

        # Define the output audio file path
        output_audio_path = "translated_audio.wav"

        # Write the concatenated translated audio to the output file
        translated_audio.write_audiofile(output_audio_path)

        # Load the original video
        video = mp.VideoFileClip(self.video_path)

        # Load the translated audio
        translated_audio = mp.AudioFileClip(output_audio_path)

        # Set the audio of the video to the translated audio
        video = video.set_audio(translated_audio)

        # Define the output video file path
        output_video_path = "translated_video.mp4"

        # Write the video with translated audio to the output file
        video.write_videofile(output_video_path, codec="libx264", audio_codec="aac")

        # Clean up temporary files
        self.cleanup_temp_files()

    def cleanup_temp_files(self):
        temp_files = ["audio.wav", "t.txt", "transcript.srt","translated_audio.wav","translated_file.srt"] + [f"translated_audio_{i}.wav" for i in range(1, 100)]  # Adjust the range accordingly
        for file in temp_files:
            if os.path.exists(file):
                os.remove(file)
                print(f"Deleted {file}")