Profiles

A profile bundles everything that decides what happens when you dictate: which hotkey starts recording, which language you speak, which models run, and how the result is formatted and inserted.

You can have several profiles. The hotkey you press decides which one runs, so a fast local profile and a high-accuracy cloud profile can sit side by side on different shortcuts.

Hotkeys

Each profile has two independent shortcuts, either of which can be left unset:

• Push to talk — recording runs while you hold the keys and stops when you release them.
• Press to start/stop — one press starts recording, another press stops it. Useful for long dictation where holding keys is uncomfortable.

The two shortcuts of a profile cannot be the same combination. Give every profile a combination that no other profile and no application you use has already claimed.

Hotkeys can be set per operating system, so the same configuration file works on a machine running macOS and one running Windows.

Language

Set the language you speak in. Leaving it unset lets the model detect the language itself, which is fine for most cases but slightly less reliable than naming it.

Language is what you speak. Translating into another language is a formatting step — see Formatting.

Modes

The mode decides where transcription runs.

Local

A Whisper model runs on your machine. Nothing leaves the device, and it works offline. You choose which Whisper model to use; larger models are more accurate and slower.

Cloud

Audio is sent to a provider you configured with your own API key — OpenAI, Groq, Fireworks, Cerebras, and others. You control the base URL, model name, and key.

Oiper

Transcription and formatting run through OiPer's hosted models. There is nothing to configure and no API key to manage. In simple mode you pick a template; in advanced mode you choose the speech model and the language model separately.

Auto submit

Auto submit presses a key combination for you right after the text is inserted, so a dictated message sends itself. The options are Disabled (the default), Enter, Ctrl + Enter, Ctrl + Shift + Enter, Shift + Enter, and Alt + Enter.

Pick the one that matches the app you dictate into most — many chat applications send on Enter, while others use Ctrl + Enter. Like hotkeys, this can be set per operating system.

Name and icon

Give each profile a name and an icon so it is recognisable in the app and in the recording overlay.