Update app.py

app.py

@@ -96,10 +96,33 @@ def one_shot(image,input_text,gender):
           f.seek(0)
           sound = AudioSegment.from_file(f.name, format="mp3")
           sound.export("/content/audio.wav", format="wav")
+
+     waveform, sample_rate = torchaudio.load(audio_in)
+     torchaudio.save("/content/audio.wav", waveform, sample_rate, encoding="PCM_S", bits_per_sample=16)
+     image = Image.open(image_in)
+     image = pad_image(image)
+     image.save("/content/image_pre.png")
+     pocketsphinx_run = subprocess.run(['pocketsphinx', '-phone_align', 'yes', 'single', '/content/audio.wav'], check=True, capture_output=True)
+     jq_run = subprocess.run(['jq', '[.w[]|{word: (.t | ascii_upcase | sub("<S>"; "sil") | sub("<SIL>"; "sil") | sub("\\\(2\\\)"; "") | sub("\\\(3\\\)"; "") | sub("\\\(4\\\)"; "") | sub("\\\[SPEECH\\\]"; "SIL") | sub("\\\[NOISE\\\]"; "SIL")), phones: [.w[]|{ph: .t | sub("\\\+SPN\\\+"; "SIL") | sub("\\\+NSN\\\+"; "SIL"), bg: (.b*100)|floor, ed: (.b*100+.d*100)|floor}]}]'], input=pocketsphinx_run.stdout, capture_output=True)
+     with open("test.json", "w") as f:
+         f.write(jq_run.stdout.decode('utf-8').strip())
+    
+     os.system(f"cd /content/one-shot-talking-face && python3 -B test_script.py --img_path /content/image_pre.png --audio_path /content/audio.wav --phoneme_path /content/test.json --save_dir /content/train")
+    
+      #1. Extract the frames from the video file using PyVideoFramesExtractor
+     os.system(f"python /content/PyVideoFramesExtractor/extract.py --video=/content/train/image_pre_audio.mp4")
+    
+      #2. Improve image quality using GFPGAN on each frames
+     os.system(f"python /content/GFPGAN/inference_gfpgan.py --upscale 2 -i /content/extracted_frames/ -o /content/video_results  --bg_upsampler realesrgan")
+    
+      #3. Merge all the frames to a one video using imageio
+     path = '/content/video_results/restored_imgs'
+     image_folder = os.fsencode(path)
+     print(image_folder)
+     filenames = []
+    
+     return  "/content/audio.wav"
        
-     result = one_shot_talking(image,'/content/audio.wav') 
-     return result
-     exit()
    elif gender == 'Male' or gender == 'male':
       print(gender)
       models, cfg, task = load_model_ensemble_and_task_from_hf_hub(