Voice dictation that doesn't send your voice to the cloud
Most dictation tools upload your audio to a server for transcription. On-device dictation instead runs the speech-recognition model on your own computer: your voice never leaves the machine, it works offline, and there's no per-minute cost. Modern local models (like Whisper) make this practical on ordinary laptops.
Cloud versus on-device: what happens to your audio
With a cloud dictation tool, your microphone audio is streamed to a company's servers, transcribed there, and sent back as text. With on-device dictation, the same work happens on your own computer. The difference matters most for the things people actually dictate: messages, notes, half-formed ideas.
Why local dictation got good
Speech models used to be too big to run outside a data center. That changed. Compact models like Whisper now transcribe accurately on an ordinary laptop, fast enough to keep up as you talk, with no account and no per-minute bill.
What to check in a dictation app's privacy claims
Read past "secure" and "encrypted" and ask the plain questions. Does it work with the network off? Does it need an account to transcribe? Where is the audio processed, and is it kept afterwards? An app that works on a plane, with no account, is telling you where your voice goes.
The honest trade-offs
On-device models are excellent, not infinite. The very largest models still live in the cloud, and a laptop from many years ago will be slower. For everyday dictation, local models are more than enough, and your voice never leaves the machine.
Where VoxClip fits
VoxClip's dictation runs on your device with a local model, so your audio never leaves it, and every dictation is saved as a searchable entry in the same Timeline as your clipboard history. It's free and needs no account. Download VoxClip free.