A free, limited version will also be available for basic needs.
Will it run on my PC?
Before you download
What you’ll need.
Organising clips takes less power than analysing hours of footage. Here’s a simple guide to choosing your setup.
Windows 10 or 11 · 64-bitWindows 10 version 1809 or newer. Windows 11 recommended.
Windows beta
Your everyday library
Organise & clip
Add recordings, create sessions, and save your favourite moments as slices.
Memory
8 GB RAM16 GB if you keep taperat open while gaming.
Processor
4 coresA 64-bit processor. 6–8 cores recommended.
Free space
1 GB + cacheKeep the library and preview cache on an SSD.
A comfortable setup: 16 GB RAM, a 6–8 core processor, and an SSD.
More context from your footage
Transcribe & analyse
Transcription makes speech searchable. Perception Engine uses diarisation and OCR to improve the text and assign speakers automatically.
Memory
16 GB RAM32 GB recommended for a comfortable analysis setup.
Processor
6 cores with AVX2AVX2 is a processor feature; check your CPU’s specifications.
Model space
2–4 GB extraAllow 2 GB for models, or 4 GB with the largest speech model.
A comfortable setup: 32 GB RAM and a processor with 8+ cores / 16 threads, plus the library’s storage and playback requirements.
Your recordings can stay put.
A hard drive, external drive, or network storage is fine for your footage. An SSD is recommended for taperat’s library and cache.
Keep your favourite recorder.
Use OBS, ShadowPlay, or your existing recording software. taperat organises the videos they create.
Start with a smaller model.
On a 4-core Windows PC, choose “base” or “small” for transcription. Larger models need more memory and take longer.
What about graphics and 4K footage?
Your graphics hardware should support playback of the format you record in, especially for 4K video. Look for hardware decoding for H.264, HEVC, or AV1 in its specifications.
Windows transcription and OCR use the processor, so an NVIDIA card or CUDA is not required. A compatible Vulkan graphics card can help with speaker separation. For the desktop app, use a 1080p display or higher; 1440p gives you more room.
Do I need an internet connection?
You’ll need it to download speech and analysis models initially; the largest speech model is about 3.1 GB. Once downloaded, transcription and OCR run locally.
For cloud-assisted analysis through MCP Bridge, you’ll also need a compatible MCP client and an internet connection. The cloud model runs with that provider; data you choose to share is handled by them.
These guidelines describe hardware needs. Actual resource use depends on your recordings and the analysis models you choose.
taperatYour library, already working for you.