read text from .txt

gfpgan/inference_gfpgan.py

@@ -96,6 +96,10 @@ def main():
         arch = 'clean'
         channel_multiplier = 2
         model_name = 'GFPGANv1.3'
+    elif args.version == '1.4':
+        arch = 'clean'
+        channel_multiplier = 2
+        model_name = 'GFPGANv1.4'
     else:
         raise ValueError(f'Wrong model version {args.version}.')
 

requirements.txt

@@ -21,4 +21,5 @@ ffmpy
 flask_ngrok2
 flask_ngrok
 opencv-python
-requests
+requests
+gradio_client

txt2video.py

@@ -20,7 +20,7 @@ def download_file(url, local_filename):
                 file.write(chunk)
 
 
-def main(job_id, video_url, text):
+def main(job_id, text, video_url):
     job_path = os.path.join(root_dir, job_id)
     os.makedirs(job_path, exist_ok=True)
 
@@ -38,7 +38,9 @@ def main(job_id, video_url, text):
     assert os.path.isfile(audio_path), f'Audio {audio_path} not exist.'
 
     wav2lip_mp4 = os.path.join(job_path, 'wav2lip.mp4')
+    print('=' * 40)
     call_wav2lip(video_path, audio_path, wav2lip_mp4)
+    print('=' * 40)
     call_gfpgan(wav2lip_mp4)
 
     output_filename = 'output.mp4'
@@ -68,14 +70,18 @@ def gen_voice(text, audio_path):
 
 if __name__ == '__main__':
     """
-    ! python run.py 2 'http://1.com.1.jpg' /content/wav2lip-gfpgan/inputs/kimk_audio.mp3
+job_id='2'
+! python txt2video.py {job_id} 'https://image.tensorartassets.com/cdn-cgi/image/plain=true,w=1280/posts/images/631302687467566953/4873c256-c1bd-49ba-85ca-5821801f122f.jpg' 海狸今天我们出去走路么
+from google.colab import files
+files.download(f'/content/jobs/{job_id}/output.mp4')
     """
     job_id = sys.argv[1]
-    video_url = sys.argv[2]
-    audio_url = sys.argv[3]
+    text = sys.argv[2]  # text or .txt file
+    video_url = sys.argv[3]
 
-    output_mp4 = main(job_id, video_url, audio_url)
+    if os.path.isfile(text):
+        # read file as string
+        with open(text, 'r') as file:
+            text = file.read()
 
-    from google.colab import files
-
-    files.download(output_mp4)
+    output_mp4 = main(job_id, text, video_url)

voice2video.py

@@ -40,7 +40,9 @@ def main(job_id, video_url, audio_url):
     assert os.path.isfile(audio_path), f'Audio {audio_path} not exist.'
 
     wav2lip_mp4 = os.path.join(job_path, 'wav2lip.mp4')
+    print('='*40)
     call_wav2lip(video_path, audio_path, wav2lip_mp4)
+    print('='*40)
     call_gfpgan(wav2lip_mp4)
 
     output_filename = 'output.mp4'

wav2lip/inference.py

@@ -225,8 +225,8 @@ def main():
     print("Number of frames available for inference: " + str(len(full_frames)))
 
     if not args.audio.endswith('.wav'):
-        print('Extracting raw audio...')
-        command = 'ffmpeg -y -i {} -strict -2 {}'.format(args.audio, f'{folder_path}/temp.wav')
+        print(f'Wav2Lip ffmpeg: Extracting raw audio {args.audio}')
+        command = 'ffmpeg -y -v error -stats -i {} -strict -2 {}'.format(args.audio, f'{folder_path}/temp.wav')
 
         subprocess.call(command, shell=True)
         args.audio = f'{folder_path}/temp.wav'
@@ -284,7 +284,8 @@ def main():
 
     out.release()
 
-    command = 'ffmpeg -y -i {} -i {} -strict -2 -q:v 1 {}'.format(args.audio, f'{folder_path}/result.avi', args.outfile)
+    print(f'Wav2Lip ffmpeg: merge into result.avi')
+    command = 'ffmpeg -y -v error -stats -i {} -i {} -strict -2 -q:v 1 {}'.format(args.audio, f'{folder_path}/result.avi', args.outfile)
     subprocess.call(command, shell=platform.system() != 'Windows')