System Requirements
ZoopZip does FREE offline speech to text entirely on your computer, so the hardware matters — bigger, more accurate models want more memory and run faster with a GPU. Here is exactly what each tier needs.
Every model is more accurate and slower than the one above it. Times are for one hour of audio and are approximate.
| Model | Download | Memory | 1 hr audio · CPU | 1 hr audio · GPU | Best for |
|---|---|---|---|---|---|
| tiny | 75 MB | ~0.4 GB | ~2 min | ~1 min | Rough, keyword-level notes |
| base | 142 MB | ~0.5 GB | ~4 min | ~1.5 min | Clear English speech |
| small recommended | 466 MB | ~1 GB | ~10 min | ~3 min | Balanced, multilingual — the default model |
| medium | 1.5 GB | ~2.6 GB | ~25 min | ~7 min | High accuracy |
| large-v3 Pro | 3.1 GB | ~4.3 GB | ~60 min | ~12 min | Best accuracy & accents, best for non-English |
Measured on Apple Silicon with Metal, ZoopZip transcribed 10 minutes of audio in about 9 seconds using the tiny model — roughly 60× real time.
macOS 12 or later, Apple Silicon or Intel. Apple Silicon Macs get automatic GPU acceleration through Metal — the fastest way to run offline speech to text on a Mac.
Windows 10 or 11, 64-bit. Runs on the CPU out of the box; NVIDIA GPU owners can enable optional CUDA acceleration for a large speed boost.
Ready?
If your computer meets the minimum above, you're good to go.
Download ZoopZip Free
ZoopZip