Update FantasyTalking/infer.py

FantasyTalking/infer.py

@@ -176,7 +176,7 @@ def main(args, pipe, fantasytalking, wav2vec_processor, wav2vec):
     duration = librosa.get_duration(filename=args.audio_path)
     print(f"🎞️ Duration: {duration:.2f}s")
 
-    num_frames = min(int(args.fps * duration // 4) * 4 + 5, args.max_num_frames)
+    num_frames = min(int(args.fps * duration), args.max_num_frames)
     print(f"📽️ Calculated number of frames: {num_frames}")
 
     print("🎧 Extracting audio features...")