Extract Subtitles 5-10x Faster with GPU Acceleration

If you have an NVIDIA GPU, you can dramatically speed up subtitle extraction in WhisperSubTranslate. This guide explains how GPU acceleration works and how to get the best performance.

CPU vs GPU Performance

The difference between CPU and GPU processing is significant. Here's a real-world comparison extracting subtitles from a 10-minute video:

Model	CPU (i7-10700)	GPU (RTX 3060)	Speedup
tiny	45 seconds	8 seconds	5.6x
small	2 minutes	15 seconds	8x
medium	6 minutes	35 seconds	10x
large-v3	15 minutes	1.5 minutes	10x

Requirements for GPU Acceleration

• NVIDIA GPU: GTX 1060 or higher recommended
• VRAM: At least 4GB for small model, 8GB+ for large models
• Drivers: Latest NVIDIA drivers installed
• CUDA: Automatically included with WhisperSubTranslate

How to Enable GPU Acceleration

1. Open WhisperSubTranslate
2. In the settings, find "Processing Device"
3. Select "GPU (CUDA)" or leave it on "Auto"
4. Start extracting subtitles - GPU will be used automatically

Optimizing GPU Performance

Choose the Right Model

Larger models provide better accuracy but require more VRAM:

• 4GB VRAM: tiny, base, small
• 6GB VRAM: medium, turbo
• 8GB+ VRAM: large-v3

Close Other Applications

For best performance, close GPU-intensive applications like games or video editors while processing.

Process Multiple Files

WhisperSubTranslate can queue multiple files. GPU processing handles batches efficiently.

Troubleshooting

GPU not detected?

• Update your NVIDIA drivers to the latest version
• Restart WhisperSubTranslate after driver update
• Check if your GPU supports CUDA (most GTX 900 series and newer)

Out of memory errors?

• Use a smaller model (small instead of medium)
• Close other GPU applications
• Process shorter video segments