Update app.py

app.py

@@ -151,8 +151,19 @@ def one_shot_talking(image_in,audio_in):
 
   # image_in_one_shot='/content/results/restored_imgs/image_pre.png'
 
-  #One Shot Talking Face algorithm
-  result=calculate(image_in,audio_in)
+  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("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")
+
   
   #Video Quality Improvement
 
@@ -163,9 +174,28 @@ def one_shot_talking(image_in,audio_in):
   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
-  merge_frames()
+  path = '/content/video_results/restored_imgs'
+  image_folder = os.fsencode(path)
+  print(image_folder)
+  filenames = []
+
+  for file in os.listdir(image_folder):
+      filename = os.fsdecode(file)
+      if filename.endswith( ('.jpg', '.png', '.gif') ):
+          filenames.append(filename)
 
-  audio_video()
+  filenames.sort() # this iteration technique has no built in order, so sort the frames
+  images = list(map(lambda filename: imageio.imread("/content/video_results/restored_imgs/"+filename), filenames))
+
+  
+  imageio.mimsave('/content/video_output.mp4', images, fps=25.0) # modify the frame duration as needed
+
+  input_video = ffmpeg.input('/content/video_output.mp4')
+
+  input_audio = ffmpeg.input('/content/audio.wav')
+
+  ffmpeg.concat(input_video, input_audio, v=1, a=1).output('final_output.mp4').run()
+  return "final_output.mp4"