wav2lip use /tmp as temp output

wav2lip/evaluation/gen_videos_from_filelist.py

@@ -164,9 +164,9 @@ def main():
 		audio_src = os.path.join(data_root, audio_src) + '.mp4'
 		video = os.path.join(data_root, video) + '.mp4'
 
-		command = 'ffmpeg -loglevel panic -y -i {} -strict -2 {}'.format(audio_src, '../temp/temp.wav')
+		command = 'ffmpeg -loglevel panic -y -i {} -strict -2 {}'.format(audio_src, '/tmp/temp.wav')
 		subprocess.call(command, shell=True)
-		temp_audio = '../temp/temp.wav'
+		temp_audio = '/tmp/temp.wav'
 
 		wav = audio.load_wav(temp_audio, 16000)
 		mel = audio.melspectrogram(wav)
@@ -208,7 +208,7 @@ def main():
 		for i, (img_batch, mel_batch, frames, coords) in enumerate(gen):
 			if i == 0:
 				frame_h, frame_w = full_frames[0].shape[:-1]
-				out = cv2.VideoWriter('../temp/result.avi', 
+				out = cv2.VideoWriter('/tmp/result.avi',
 								cv2.VideoWriter_fourcc(*'DIVX'), fps, (frame_w, frame_h))
 
 			img_batch = torch.FloatTensor(np.transpose(img_batch, (0, 3, 1, 2))).to(device)
@@ -231,7 +231,7 @@ def main():
 		vid = os.path.join(args.results_dir, '{}.mp4'.format(idx))
 
 		command = 'ffmpeg -loglevel panic -y -i {} -i {} -strict -2 -q:v 1 {}'.format(temp_audio, 
-								'../temp/result.avi', vid)
+								'/tmp/result.avi', vid)
 		subprocess.call(command, shell=True)
 
 if __name__ == '__main__':

wav2lip/evaluation/real_videos_inference.py

@@ -214,9 +214,9 @@ def main():
 		audio_src = os.path.join(args.data_root, audio_src)
 		video = os.path.join(args.data_root, video)
 
-		command = 'ffmpeg -loglevel panic -y -i {} -strict -2 {}'.format(audio_src, '../temp/temp.wav')
+		command = 'ffmpeg -loglevel panic -y -i {} -strict -2 {}'.format(audio_src, '/tmp/temp.wav')
 		subprocess.call(command, shell=True)
-		temp_audio = '../temp/temp.wav'
+		temp_audio = '/tmp/temp.wav'
 
 		wav = audio.load_wav(temp_audio, 16000)
 		mel = audio.melspectrogram(wav)
@@ -275,7 +275,7 @@ def main():
 			if i == 0:
 				frame_h, frame_w = full_frames[0].shape[:-1]
 
-				out = cv2.VideoWriter('../temp/result.avi', 
+				out = cv2.VideoWriter('/tmp/result.avi',
 								cv2.VideoWriter_fourcc(*'DIVX'), fps, (frame_w, frame_h))
 
 			img_batch = torch.FloatTensor(np.transpose(img_batch, (0, 3, 1, 2))).to(device)
@@ -296,8 +296,8 @@ def main():
 		out.release()
 
 		vid = os.path.join(args.results_dir, '{}.mp4'.format(idx))
-		command = 'ffmpeg -loglevel panic -y -i {} -i {} -strict -2 -q:v 1 {}'.format('../temp/temp.wav', 
-								'../temp/result.avi', vid)
+		command = 'ffmpeg -loglevel panic -y -i {} -i {} -strict -2 -q:v 1 {}'.format('/tmp/temp.wav',
+								'/tmp/result.avi', vid)
 		subprocess.call(command, shell=True)
 
 

wav2lip/inference.py

@@ -253,7 +253,7 @@ def main():
 			print ("Model loaded")
 
 			frame_h, frame_w = full_frames[0].shape[:-1]
-			out = cv2.VideoWriter('temp/result.avi', 
+			out = cv2.VideoWriter('/tmp/result.avi',
 									cv2.VideoWriter_fourcc(*'DIVX'), fps, (frame_w, frame_h))
 
 		img_batch = torch.FloatTensor(np.transpose(img_batch, (0, 3, 1, 2))).to(device)
@@ -273,7 +273,7 @@ def main():
 
 	out.release()
 
-	command = 'ffmpeg -y -i {} -i {} -strict -2 -q:v 1 {}'.format(args.audio, 'temp/result.avi', args.outfile)
+	command = 'ffmpeg -y -i {} -i {} -strict -2 -q:v 1 {}'.format(args.audio, '/tmp/result.avi', args.outfile)
 	subprocess.call(command, shell=platform.system() != 'Windows')
 
 if __name__ == '__main__':