Update app.py

app.py

@@ -29,9 +29,9 @@ def get_video_res(img_path, audio_path, res_video_path, dynamic_scale=1.0):
     min_resolution = 512
     inference_steps = 25
 
-    # Get audio duration (정보 출력용)
+    # Get audio duration (for logging)
     audio = AudioSegment.from_file(audio_path)
-    duration = len(audio) / 1000.0  # 초 단위 변환
+    duration = len(audio) / 1000.0  # Convert ms to seconds
 
     face_info = pipe.preprocess(img_path, expand_ratio=expand_ratio)
     print(f"Face detection info: {face_info}")
@@ -43,7 +43,7 @@ def get_video_res(img_path, audio_path, res_video_path, dynamic_scale=1.0):
         img_path = crop_image_path
         os.makedirs(os.path.dirname(res_video_path), exist_ok=True)
 
-        # NOTE: Sonic.process()는 더 이상 duration 인자를 받지 않으므로 제거합니다.
+        # Process the video (duration parameter removed)
         pipe.process(
             img_path, 
             audio_path, 
@@ -52,6 +52,8 @@ def get_video_res(img_path, audio_path, res_video_path, dynamic_scale=1.0):
             inference_steps=inference_steps,
             dynamic_scale=dynamic_scale
         )
+        # ★ 수정: 생성된 비디오 파일 경로를 반환하도록 함.
+        return res_video_path
     else:
         return -1
 
@@ -61,7 +63,7 @@ os.makedirs(tmp_path, exist_ok=True)
 os.makedirs(res_path, exist_ok=True)
 
 def process_sonic(image, audio, dynamic_scale):
-    # 입력 검증
+    # Input validation
     if image is None:
         raise gr.Error("Please upload an image")
     if audio is None:
@@ -75,7 +77,7 @@ def process_sonic(image, audio, dynamic_scale):
     if len(arr.shape) == 1:
         arr = arr[:, None]
     
-    # 오디오 세그먼트 생성
+    # Create an audio segment from numpy array
     audio_segment = AudioSegment(
         arr.tobytes(),
         frame_rate=sampling_rate,
@@ -84,18 +86,18 @@ def process_sonic(image, audio, dynamic_scale):
     )
     audio_segment = audio_segment.set_frame_rate(sampling_rate)
     
-    # 파일 경로 생성
+    # Generate file paths
     image_path = os.path.abspath(os.path.join(tmp_path, f'{img_md5}.png'))
     audio_path = os.path.abspath(os.path.join(tmp_path, f'{audio_md5}.wav'))
     res_video_path = os.path.abspath(os.path.join(res_path, f'{img_md5}_{audio_md5}_{dynamic_scale}.mp4'))
     
-    # 입력 파일이 없으면 저장
+    # Save input files if they don't exist
     if not os.path.exists(image_path):
         image.save(image_path)
     if not os.path.exists(audio_path):
         audio_segment.export(audio_path, format="wav")
     
-    # 캐시된 결과가 있으면 반환, 없으면 새로 생성
+    # If cached video exists, return it; otherwise, generate a new one
     if os.path.exists(res_video_path):
         print(f"Using cached result: {res_video_path}")
         return res_video_path
@@ -103,7 +105,7 @@ def process_sonic(image, audio, dynamic_scale):
         print(f"Generating new video with dynamic scale: {dynamic_scale}")
         return get_video_res(image_path, audio_path, res_video_path, dynamic_scale)
 
-# 예시 데이터를 위한 dummy 함수 (필요에 따라 실제 예시 데이터를 넣으세요)
+# Dummy get_example function to prevent errors in examples section
 def get_example():
     return []
 
@@ -171,7 +173,7 @@ with gr.Blocks(css=css) as demo:
                 elem_id="video_output"
             )
     
-    # 버튼 클릭 시 애니메이션 생성 함수 호출
+    # Process button click: when clicked, process_sonic() is called and its return value is sent to video_output.
     process_btn.click(
         fn=process_sonic,
         inputs=[image_input, audio_input, dynamic_scale],
@@ -179,7 +181,7 @@ with gr.Blocks(css=css) as demo:
         api_name="animate"
     )
     
-    # 예시 섹션
+    # Examples section
     gr.Examples(
         examples=get_example(),
         fn=process_sonic,
@@ -188,7 +190,7 @@ with gr.Blocks(css=css) as demo:
         cache_examples=False
     )
     
-    # Footer: Attribution & Links
+    # Footer with attribution and links
     gr.HTML("""
         <div style="text-align: center; margin-top: 2em;">
             <div style="margin-bottom: 1em;">
@@ -203,5 +205,5 @@ with gr.Blocks(css=css) as demo:
         </div>
     """)
 
-# 공유 링크 생성: share=True
+# To create a public link, share=True is set.
 demo.launch(share=True)