GPU and VRAM
Reference-voice Live translation can fail or slow down when VRAM is insufficient. Check the VRAM Counter and Model Status in the app.
VoxShift is Windows voice translation software. Live translation with reference-voice speech generation depends on GPU, VRAM, audio devices, selected models, and your routing setup.
| Area | Requirement | Notes |
|---|---|---|
| OS | Windows 10 version 2004 or later / Windows 11 | x64 PC with desktop UI support |
| GPU / VRAM | GPU acceleration recommended | Required VRAM depends on selected speech recognition, translation, and voice generation models |
| Audio | Microphone, output device, monitor output | Virtual audio may be needed for Discord or game VC routing |
| Storage | Local model cache | Free disk space depends on selected models and logs |
Reference-voice Live translation can fail or slow down when VRAM is insufficient. Check the VRAM Counter and Model Status in the app.
If microphone, main output, monitor output, or virtual audio routing is wrong, Discord or OBS will not receive the translated speech.
Download, cache, and loading state affect first-run setup time and live processing speed.
Games and streaming software can consume GPU/CPU resources, which affects VoxShift latency.
Requirements cannot be judged from a table alone. Test your own PC, audio devices, and target app before deciding to purchase.
Requirements vary by selected model, language pair, quality preset, and other apps running at the same time. Use the Free Trial Readiness Check on your own PC.
Low Spec Mode exists, but reference-voice Live translation still depends on GPU and VRAM. Test latency and quality before buying.
If you want to send translated speech into Discord or game voice chat as a microphone input, a virtual audio device may be needed. The Microsoft Store build does not bundle third-party driver installers.