Quickstart

The first time you open OiPer it walks you through three steps and creates your first profile from your answers.

1. Choose how to transcribe

Pick one of three modes:

• Oiper — OiPer hosts and updates the models. Nothing to configure.
• Local — a Whisper model runs on your machine, fully offline.
• Cloud — your audio is sent to a provider such as OpenAI or Groq using your own API key.

If you pick Local, choose a Whisper model here as well and select Download & Continue so the model is fetched while you finish setup. Continue without downloading also works; you can download it later from the Models page.

If you pick Cloud, add an ASR model with its provider, base URL, model name, and API key.

2. Choose text formatting

Formatting sends the raw transcript through a language model to remove filler words and fix punctuation and capitalization, turning um so i need the report by friday ok into I need the report by Friday.

Turn the switch on to enable it. In Oiper mode this needs no further setup. In local or cloud mode, formatting runs through an LLM, so add one now or skip and configure it later from the Models page.

3. Set your shortcut

Record the hotkey you will hold while speaking. The default is Left Shift + Left Control. Pick a combination you do not already use in other applications.

Select Let's be productive to finish. Your profile is saved and the main window opens.

Your first dictation

1. Click into any text field in any application.
2. Hold your hotkey.
3. Say a sentence.
4. Release the hotkey.

The transcript is inserted where your cursor was. Every run is recorded on the History page, so if nothing appears you can check there to see whether the problem was recording, transcription, or insertion. The Troubleshooting page covers the common causes.