Guide
Local recognition vs cloud dictation.
Cloud apps send microphone audio to a remote model. Nimva runs recognition on your Mac instead. Here are clear, claim-safe tradeoffs.
Last updated: 2026-10-02
Local vs cloud dictation: cloud tools typically send audio (or derived features) to a remote model; Nimva keeps recognition on your Mac. Local means no Nimva word meter for speech recognition, with a speech model installed on disk.
Side-by-side
| Topic | Local (Nimva) | Typical cloud dictation |
|---|---|---|
| Where recognition runs | On your Mac | Remote service |
| Audio for recognition | Not uploaded to Nimva | Often streamed or uploaded |
| Word / minute meter | No meter for local STT | Often metered or tiered |
| Offline after model install | Yes (license may still check later) | Usually needs network |
| Tradeoff | Download/install speech model | Less local disk; depends on vendor privacy |
When local fits
Choose local when you want dictation audio to stay on the device for recognition and you accept installing a speech model. For competitor matrices (prices change—verify before you buy), see Compare.
More detail: On-device Mac dictation.