<Bug summary>

[dangnilaole]

Bug Description

When upscaling a long video (e.g., ~20 minutes), the output video suffers from severe audio desynchronization and the video appears to speed up significantly in the later part. The first few minutes are usually in sync, but around the halfway mark the video starts to drift, running faster than the audio, making the result unusable.

Environment

• video2x version: (e.g. 4.8.1 / commit hash)
• Operating System: (e.g. Windows 11 23H2 / Ubuntu 22.04)
• Python version: (e.g. 3.10.12)
• GPU / Driver: (e.g. NVIDIA RTX 4060, driver 551.86, CUDA 12.4)
• Installation method: (pip / portable / built from source)
• Upscaler used: (waifu2x-ncnn-vulkan / realcugan / srmd / …)
• FFmpeg version: (e.g. 6.1.1)

Steps to Reproduce

1. Take a video file about 20 minutes long (e.g. 1920×1080, 30 fps, H.264 video, AAC audio).
2. Run a command like:

   video2x -i input.mp4 -o output.mp4 -p3 --width 3840 --height 2160