Help & support.
Direct answers about recording, transcription, and what leaves your device.
Three steps.
setup · record · refineSet up.
The first time you open OhmsVoice, a setup card asks how it should listen: built-in recognition, or the open speech model, a one-time download you approve. Nothing is recorded until one of them is ready.
Record.
On a Mac, hold the system-wide shortcut (default Control-Option-Space), use the menu bar item, or press Command-R. On iPhone and iPad, use the Record tab. The text lands on the clipboard, ready to paste.
Refine.
Snippets and spellings in Settings work on the device with no key. Optional rewrite presets use a provider you connect with your own key, or on a Mac a model server running on that Mac.
Your keys, your device.
byok · privacy · pricingSo your first recording is never thrown away. Until a speech engine is ready, the microphone stays closed and a setup card offers two choices: built-in recognition, ready at once where your device supports your language, or the open speech model, a one-time 483 MB download you approve, checksummed and load-tested before use. You can switch later in Settings under Dictation.
No. Transcription, Raw Transcript, Clean (On-Device), snippets and saved spellings all work with no key and no network once a speech engine is set up. A key is only needed for the optional rewrite presets (Clean & Concise, Executive Email, Developer Bug / Jira Ticket), billed by that provider directly to you. On a Mac you can use a model server running on that Mac instead of a key.
The open speech model transcribes 25 languages with automatic punctuation. Built-in recognition works wherever your device offers on-device recognition for your language, and the setup card tells you if it does not. Filler removal and spoken commands such as “new line” in Clean (On-Device) understand English only. The app itself is in English.
Audio never does. Apart from links you open, the network is used only if you download the open speech model or run a rewrite preset: then the transcript, the preset's instructions and any spellings you saved go straight from your device to the provider you approved, under your key. Through version 1.1.2, a rewrite started on a Mac also includes the name of the app you were dictating into. Silo Systems runs no server and receives no copy.
In the system Keychain, in app-scoped items. Never in preferences, files, or logs, and never transmitted to Silo Systems.
Settings lists the built-in cloud providers that accept your own API key, and the privacy policy names each one. On a Mac you can also use a model server running on that Mac. Saving a key makes that provider active, and it stays active across launches. You pay the provider directly; OhmsVoice adds no markup.
A one-time purchase at $29.99 covering Mac, iPhone, and iPad. No subscription, no account.