import os import google.generativeai as genai from moviepy.video.io.VideoFileClip import VideoFileClip from moviepy.audio.io.AudioFileClip import AudioFileClip from moviepy.video.compositing.CompositeVideoClip import CompositeVideoClip from moviepy.video.tools.subtitles import SubtitlesClip import tempfile import logging import gradio as gr from gtts import gTTS import srt # Suppress moviepy logs logging.getLogger("moviepy").setLevel(logging.ERROR) # Configure Gemini API genai.configure(api_key=os.environ["GEMINI_API_KEY"]) # Create the Gemini model generation_config = { "temperature": 0.7, "top_p": 0.9, "top_k": 40, "max_output_tokens": 8192, "response_mime_type": "text/plain", } model = genai.GenerativeModel( model_name="gemini-2.5-flash-preview-05-20", generation_config=generation_config, ) # List of all supported languages SUPPORTED_LANGUAGES = [ "Auto Detect", "English", "Chinese", "German", "Spanish", "Russian", "Korean", "French", "Japanese", "Portuguese", "Turkish", "Polish", "Catalan", "Dutch", "Arabic", "Swedish", "Italian", "Indonesian", "Hindi", "Finnish", "Vietnamese", "Hebrew", "Ukrainian", "Greek", "Malay", "Czech", "Romanian", "Danish", "Hungarian", "Tamil", "Norwegian", "Thai", "Urdu", "Croatian", "Bulgarian", "Lithuanian", "Latin", "Maori", "Malayalam", "Welsh", "Slovak", "Telugu", "Persian", "Latvian", "Bengali", "Serbian", "Azerbaijani", "Slovenian", "Kannada", "Estonian", "Macedonian", "Breton", "Basque", "Icelandic", "Armenian", "Nepali", "Mongolian", "Bosnian", "Kazakh", "Albanian", "Swahili", "Galician", "Marathi", "Punjabi", "Sinhala", "Khmer", "Shona", "Yoruba", "Somali", "Afrikaans", "Occitan", "Georgian", "Belarusian", "Tajik", "Sindhi", "Gujarati", "Amharic", "Yiddish", "Lao", "Uzbek", "Faroese", "Haitian Creole", "Pashto", "Turkmen", "Nynorsk", "Maltese", "Sanskrit", "Luxembourgish", "Burmese", "Tibetan", "Tagalog", "Malagasy", "Assamese", "Tatar", "Hawaiian", "Lingala", "Hausa", "Bashkir", "Javanese", "Sundanese" ] # Language code mapping for gTTS LANGUAGE_CODES = { "English": "en", "Chinese": "zh", "German": "de", "Spanish": "es", "Russian": "ru", "Korean": "ko", "French": "fr", "Japanese": "ja", "Portuguese": "pt", "Turkish": "tr", "Polish": "pl", "Catalan": "ca", "Dutch": "nl", "Arabic": "ar", "Swedish": "sv", "Italian": "it", "Indonesian": "id", "Hindi": "hi", "Finnish": "fi", "Vietnamese": "vi", "Hebrew": "he", "Ukrainian": "uk", "Greek": "el", "Malay": "ms", "Czech": "cs", "Romanian": "ro", "Danish": "da", "Hungarian": "hu", "Tamil": "ta", "Norwegian": "no", "Thai": "th", "Urdu": "ur", "Croatian": "hr", "Bulgarian": "bg", "Lithuanian": "lt", "Latin": "la", "Maori": "mi", "Malayalam": "ml", "Welsh": "cy", "Slovak": "sk", "Telugu": "te", "Persian": "fa", "Latvian": "lv", "Bengali": "bn", "Serbian": "sr", "Azerbaijani": "az", "Slovenian": "sl", "Kannada": "kn", "Estonian": "et", "Macedonian": "mk", "Breton": "br", "Basque": "eu", "Icelandic": "is", "Armenian": "hy", "Nepali": "ne", "Mongolian": "mn", "Bosnian": "bs", "Kazakh": "kk", "Albanian": "sq", "Swahili": "sw", "Galician": "gl", "Marathi": "mr", "Punjabi": "pa", "Sinhala": "si", "Khmer": "km", "Shona": "sn", "Yoruba": "yo", "Somali": "so", "Afrikaans": "af", "Occitan": "oc", "Georgian": "ka", "Belarusian": "be", "Tajik": "tg", "Sindhi": "sd", "Gujarati": "gu", "Amharic": "am", "Yiddish": "yi", "Lao": "lo", "Uzbek": "uz", "Faroese": "fo", "Haitian Creole": "ht", "Pashto": "ps", "Turkmen": "tk", "Nynorsk": "nn", "Maltese": "mt", "Sanskrit": "sa", "Luxembourgish": "lb", "Burmese": "my", "Tibetan": "bo", "Tagalog": "tl", "Malagasy": "mg", "Assamese": "as", "Tatar": "tt", "Hawaiian": "haw", "Lingala": "ln", "Hausa": "ha", "Bashkir": "ba", "Javanese": "jv", "Sundanese": "su" } def extract_audio_from_video(video_file): """Extract audio from a video file and save it as a WAV file.""" video = VideoFileClip(video_file) audio_file = os.path.join(tempfile.gettempdir(), "extracted_audio.wav") video.audio.write_audiofile(audio_file, fps=16000, logger=None) return audio_file def transcribe_audio_with_gemini(audio_file): """Transcribe audio using Gemini with a magic prompt for accurate timestamps.""" with open(audio_file, "rb") as f: audio_data = f.read() audio_blob = { 'mime_type': 'audio/wav', 'data': audio_data } prompt = """ You are a professional transcriber. Transcribe this audio accurately and verbatim in the original language. Include timestamps for each sentence in the following format: [HH:MM:SS] Sentence 1 [HH:MM:SS] Sentence 2 ... Ensure the timestamps are accurate and correspond to the start of each sentence. Respond only with the transcription and timestamps. Do not add explanations or extra text. """ convo = model.start_chat() convo.send_message(prompt) response = convo.send_message(audio_blob) return response.text.strip() def generate_subtitles(transcription): """Generate SRT subtitles from transcription with timestamps.""" lines = transcription.split("\n") srt_subtitles = [] for i, line in enumerate(lines, start=1): if not line.strip(): continue if line.startswith("["): timestamp = line.split("]")[0] + "]" text = line.split("]")[1].strip() else: timestamp = "[00:00:00]" text = line.strip() start_time = timestamp[1:-1] start_seconds = time_to_seconds(start_time) end_seconds = start_seconds + 5 # Placeholder duration subtitle = srt.Subtitle( index=i, start=datetime.timedelta(seconds=start_seconds), end=datetime.timedelta(seconds=end_seconds), content=text ) srt_subtitles.append(subtitle) return srt.compose(srt_subtitles) def time_to_seconds(time_str): """Convert HH:MM:SS to seconds.""" hh, mm, ss = map(int, time_str.split(":")) return hh * 3600 + mm * 60 + ss def seconds_to_time(seconds): """Convert seconds to HH:MM:SS.""" hh = seconds // 3600 mm = (seconds % 3600) // 60 ss = seconds % 60 return f"{hh:02}:{mm:02}:{ss:02}" def translate_srt(srt_text, target_language): """Translate an SRT file while preserving timestamps.""" prompt = f""" Translate the following SRT subtitles into {target_language}. Preserve the SRT format (timestamps and structure). Translate only the text after the timestamp. Do not add explanations or extra text. Ensure the translation is accurate and culturally appropriate. Here is the SRT file: {srt_text} """ response = model.generate_content(prompt) return response.text def generate_tts_audio(srt_text, language): """Generate TTS audio from SRT text.""" # Extract all text from SRT subtitles = list(srt.parse(srt_text)) all_text = " ".join([sub.content for sub in subtitles]) # Get language code lang_code = LANGUAGE_CODES.get(language, "en") # Generate TTS tts = gTTS(text=all_text, lang=lang_code, slow=False) audio_file = os.path.join(tempfile.gettempdir(), "tts_audio.mp3") tts.save(audio_file) return audio_file def add_subtitles_to_video(video_file, srt_file, output_file): """Add subtitles to video and return the path to the new video.""" # Create subtitle clip generator = lambda txt: TextClip(txt, font='Arial', fontsize=24, color='white') subtitles = SubtitlesClip(srt_file, generator) # Load video video = VideoFileClip(video_file) # Composite video with subtitles result = CompositeVideoClip([ video, subtitles.set_position(('center', 'bottom')) ]) # Write output result.write_videofile(output_file, codec='libx264', audio_codec='aac', threads=4) return output_file def process_video(video_file, language="Auto Detect", translate_to=None, add_tts=False, add_subtitles=False): """Process a video file with full options.""" # Extract audio from the video audio_file = extract_audio_from_video(video_file) # Transcribe audio using Gemini transcription = transcribe_audio_with_gemini(audio_file) # Generate subtitles subtitles = generate_subtitles(transcription) # Save original subtitles original_srt_file = os.path.join(tempfile.gettempdir(), "original_subtitles.srt") with open(original_srt_file, "w", encoding="utf-8") as f: f.write(subtitles) # Translate subtitles if requested translated_srt_file = None if translate_to and translate_to != "None": translated_subtitles = translate_srt(subtitles, translate_to) translated_srt_file = os.path.join(tempfile.gettempdir(), "translated_subtitles.srt") with open(translated_srt_file, "w", encoding="utf-8") as f: f.write(translated_subtitles) # Generate TTS audio if requested tts_audio_file = None if add_tts: target_lang = translate_to if translate_to and translate_to != "None" else language tts_audio_file = generate_tts_audio(subtitles if not translated_srt_file else translated_subtitles, target_lang) # Create video with subtitles if requested output_video_file = None if add_subtitles: srt_to_use = translated_srt_file if translated_srt_file else original_srt_file output_video_file = os.path.join(tempfile.gettempdir(), "output_video.mp4") add_subtitles_to_video(video_file, srt_to_use, output_video_file) # Clean up os.remove(audio_file) return original_srt_file, translated_srt_file, tts_audio_file, output_video_file, "Detected Language: Auto" # Define the Gradio interface with gr.Blocks(title="AutoSubGen Pro - AI Video Subtitle Generator") as demo: # Header with gr.Column(): gr.Markdown("# 🎥 AutoSubGen Pro") gr.Markdown("### Advanced AI-Powered Video Subtitle Generator") gr.Markdown("Generate, translate, and add subtitles with text-to-speech audio to your videos.") # Main content with gr.Tab("Generate Subtitles"): gr.Markdown("### Upload a video file to process") with gr.Row(): video_input = gr.Video(label="Upload Video File", scale=2) with gr.Column(): language_dropdown = gr.Dropdown( choices=SUPPORTED_LANGUAGES, label="Source Language", value="Auto Detect", ) translate_to_dropdown = gr.Dropdown( choices=["None"] + SUPPORTED_LANGUAGES[1:], label="Translate To", value="None", ) tts_checkbox = gr.Checkbox(label="Generate Text-to-Speech Audio") subtitles_checkbox = gr.Checkbox(label="Add Subtitles to Video") generate_button = gr.Button("Process Video", variant="primary") with gr.Row(): with gr.Column(): original_subtitle_output = gr.File(label="Original Subtitles (SRT)") translated_subtitle_output = gr.File(label="Translated Subtitles (SRT)") with gr.Column(): tts_audio_output = gr.Audio(label="Text-to-Speech Audio", visible=False) video_output = gr.Video(label="Video with Subtitles", visible=False) detected_language_output = gr.Textbox(label="Detected Language") # Show/hide outputs based on checkboxes def toggle_outputs(tts, subs): return [ gr.Audio(visible=tts), gr.Video(visible=subs) ] tts_checkbox.change( fn=lambda x: gr.Audio(visible=x), inputs=tts_checkbox, outputs=tts_audio_output ) subtitles_checkbox.change( fn=lambda x: gr.Video(visible=x), inputs=subtitles_checkbox, outputs=video_output ) # Link button to function generate_button.click( process_video, inputs=[video_input, language_dropdown, translate_to_dropdown, tts_checkbox, subtitles_checkbox], outputs=[original_subtitle_output, translated_subtitle_output, tts_audio_output, video_output, detected_language_output] ) # Launch the interface demo.launch(share=True)