Troubleshooting

If a GIMZWARE plugin doesn't appear in your plugin list:

• Rescan plugins - Most DAWs have a "Rescan" option in plugin settings
• Check plugin format - Ensure you're looking in VST3 (not VST2) or AU
• Verify installation path - Check that files exist in the correct directory
• 64-bit only - All GIMZWARE plugins require a 64-bit DAW

If you can't activate your licence:

• Check internet connection - Initial activation requires internet
• Verify licence key - Copy/paste from email to avoid typos
• Check activation limit - Licence allows 2 machines; deactivate old ones if needed
• Firewall - Ensure gimzware.ai is not blocked

Contact [email protected] if issues persist.

If you experience crackling, pops, or dropouts:

• Increase buffer size - Try 512 or 1024 samples
• Enable GPU acceleration - Significantly reduces CPU load
• Close other applications - Free up system resources
• Update GPU drivers - Especially for CUDA users
• Disable power saving - Use "High Performance" power plan

If CPU usage is higher than expected:

• Enable GPU acceleration - Primary solution for CPU load
• Check execution provider - in Settings, verify the GPU is selected
• Reduce instances - Use one StemX per session if possible
• Use File mode - Pre-process files for lower live CPU usage

If StemX doesn't detect your GPU:

• NVIDIA users - Update to driver 525+ for CUDA support
• Windows users - Ensure DirectX 12 is installed
• Check GPU compatibility - GTX 1060 or newer required
• Restart DAW - Required after driver updates

If stems contain artifacts or bleed:

• Source quality matters - Higher quality input = better separation
• Complex arrangements - Dense mixes are more challenging
• Avoid double-processing - Don't separate already-separated stems
• Use File mode for critical work - Slightly better quality than Live mode

If CPU usage spikes when using high oversampling rates:

• Reduce oversampling - 4x or 8x delivers excellent quality at a fraction of the CPU cost
• Use Live quality mode - The Live preset uses lower oversampling optimised for real-time tracking
• Bounce/freeze tracks - Render tracks that don't need further editing to free up CPU headroom
• Increase buffer size - Larger buffers reduce the relative overhead of oversampled processing

Neural amp models are input-gain sensitive, just like real hardware:

• Adjust Drive and Input - The character of neural models changes with drive level; experiment with both controls to find the sweet spot
• Match source levels - Guitar and bass signals vary widely; calibrate Input so peaks sit around −12 dBFS before the drive stage
• Use the LowCut filter - Rolling off sub-bass before distortion reduces muddiness and gives cleaner response from neural models
• Check oversampling - Higher oversampling (8x or 16x) can affect the perceived tone; try different rates when dialling in neural tones