Download app.py from konay233/Recap: direct link, hf CLI and curl.
- Browser
- Download file 5.02 kB
-
https://huggingface.co/spaces/konay233/Recap/resolve/main/app.py
- Command line
-
hf download hf://spaces/konay233/Recap/app.py
-
curl -L -o app.py https://huggingface.co/spaces/konay233/Recap/resolve/main/app.py
5.02 kB
| import os | |
| import json | |
| import asyncio | |
| import subprocess | |
| import gradio as gr | |
| import assemblyai as aai | |
| import edge_tts | |
| from static_ffmpeg import add_paths | |
| # Static FFmpeg Setup | |
| add_paths() | |
| # AssemblyAI Key | |
| ASSEMBLYAI_API_KEY = "0cc0d50376e5421da2b362efca91725a" | |
| aai.settings.api_key = ASSEMBLYAI_API_KEY | |
| def process_video(video_path): | |
| if not video_path or not os.path.exists(video_path): | |
| return "Video file တင်ပေးပါ သို့မဟုတ် ဖိုင်မတွေ့ပါ။", "" | |
| try: | |
| transcriber = aai.Transcriber() | |
| transcript = transcriber.transcribe(video_path) | |
| formatted_inputs = [] | |
| for idx, sentence in enumerate(transcript.get_sentences()): | |
| start_sec = round(sentence.start / 1000, 1) | |
| end_sec = round(sentence.end / 1000, 1) | |
| max_limit = round(end_sec - start_sec, 1) | |
| formatted_inputs.append(f"ID: {idx+1} | Time Frame: {start_sec}s to {end_sec}s (Max Limit: {max_limit}s) | Text: \"{sentence.text}\"") | |
| prompt_header = ( | |
| "You are a professional Burmese dubbing adapter. Translate each English segment into natural, spoken Burmese for voice-over. " | |
| "Preserve the meaning, emotion, names, and order. Return ONLY valid JSON in this exact shape: {\"translations\":[\"...\"]}. " | |
| "The array must have one string for every segment, in matching ID order. Keep each translation concise enough to fit its time frame and never add commentary.\n\n" | |
| "Translate these timestamped segments:\n" | |
| ) | |
| full_prompt = prompt_header + "\n".join(formatted_inputs) | |
| return full_prompt, "Subtitle Extraction အောင်မြင်ပါပြီ။ အထက်ပါ Prompt ကို Gemini App တွင် ဘာသာပြန်ပါ။" | |
| except Exception as e: | |
| return f"Error ဖြစ်သွားပါသည်: {str(e)}", "AssemblyAI API Key မှန်မမှန် ပြန်စစ်ပါ။" | |
| async def generate_dubbed_video_async(video_path, json_str, voice_option): | |
| try: | |
| data = json.loads(json_str) | |
| translations = data.get("translations", []) | |
| except Exception as e: | |
| return None, f"JSON Format မှားယွင်းနေပါသည်: {str(e)}" | |
| selected_voice = "my-MM-ThihaNeural" if "Thiha" in voice_option else "my-MM-NilarNeural" | |
| transcriber = aai.Transcriber() | |
| transcript = transcriber.transcribe(video_path) | |
| sentences = list(transcript.get_sentences()) | |
| if len(translations) != len(sentences): | |
| return None, f"စာကြောင်း အရေအတွက် မတူပါ (Original: {len(sentences)}, Gemini: {len(translations)})" | |
| audio_files = [] | |
| for idx, (sentence, burmese_text) in enumerate(zip(sentences, translations)): | |
| audio_path = f"seg_{idx}.mp3" | |
| communicate = edge_tts.Communicate(burmese_text, selected_voice) | |
| await communicate.save(audio_path) | |
| audio_files.append(audio_path) | |
| output_video = "dubbed_output.mp4" | |
| cmd = f"ffmpeg -y -i \"{video_path}\" -i \"{audio_files[0]}\" -c:v copy -c:a aac -map 0:v:0 -map 1:a:0 \"{output_video}\"" | |
| subprocess.run(cmd, shell=True) | |
| return output_video, f"Video Dubbing ပြီးပါပြီ။ (အသုံးပြုထားသော အသံ - {voice_option})" | |
| def generate_dubbed_video(video_path, json_str, voice_option): | |
| return asyncio.run(generate_dubbed_video_async(video_path, json_str, voice_option)) | |
| with gr.Blocks(title="Movie Recap Dubbing Tool") as demo: | |
| gr.Markdown("## 🎬 Semi-Manual Video Dubbing System (Free Space)") | |
| video_input = gr.Video(label="Upload Video File") | |
| extract_btn = gr.Button("၁။ Subtitles & Prompt ထုတ်ယူရန်", variant="primary") | |
| # show_copy_button ကို ဖြုတ်ပေးထားပါသည် | |
| prompt_output = gr.Textbox(label="Gemini App ထဲ Copy ကူးထည့်ရမည့် Prompt Block", lines=10) | |
| status_1 = gr.Textbox(label="Status") | |
| extract_btn.click(process_video, inputs=[video_input], outputs=[prompt_output, status_1]) | |
| gr.Markdown("---") | |
| with gr.Row(): | |
| json_input = gr.Textbox(label="၂။ Gemini App မှ ရလာသော JSON Output Paste ထည့်ရန်", lines=6) | |
| voice_dropdown = gr.Dropdown(choices=["အမျိုးသားအသံ (Thiha)", "အမျိုးသမီးအသံ (Nilar)"], value="အမျိုးသားအသံ (Thiha)", label="TTS အသံ ရွေးရန်") | |
| render_btn = gr.Button("၃။ မြန်မာ အသံသွင်းပြီး Video ထုတ်ယူရန်", variant="primary") | |
| video_output = gr.Video(label="Final Dubbed Video") | |
| status_2 = gr.Textbox(label="Status") | |
| render_btn.click(generate_dubbed_video, inputs=[video_input, json_input, voice_dropdown], outputs=[video_output, status_2]) | |
| demo.launch(server_name="0.0.0.0") | |