OiPer
DownloadDownloadDownloadDownloadDownloadDownload

A Whisper desktop app for Windows, Mac and Linux

Whisper is OpenAI’s speech recognition model, and running it usually means Python, a command line and a pile of dependencies. OiPer packages local Whisper into a desktop app: pick a model, download it once, then hold a hotkey and speak. The text is typed into whatever app you are using, and it works offline.

Updated October 12, 2026

Download OiPerDownload for WindowsDownload for macOSDownload for LinuxDownload for LinuxDownload for LinuxRead the docs

Whisper without the setup

An installer, a model picker and a hotkey. No Python and no terminal.

Install like any other app

Download the Windows installer, the Apple Silicon or Intel build for macOS 10.15 and later, or the AppImage, .deb or .rpm for Linux.

Read about Installation

Models download inside the app

Choose a Whisper model during setup or on the Models page. It is downloaded once from Hugging Face and then runs entirely on your machine.

Read about Models

Uses your GPU automatically

The Whisper Hardware setting defaults to Auto, which uses the GPU when one is usable and the CPU otherwise. Force CPU if GPU transcription misbehaves.

Read about Settings

Types into any app

Hold your hotkey, speak and release. The transcript is inserted at your cursor in your editor, browser, email, chat or terminal.

Read about Quickstart

How to choose a Whisper model

Larger models are more accurate but slower and need more disk space and memory.

Start with Base (142 MB)

A good balance for everyday dictation and the recommended starting point on most computers.

Read about Models

Step up to Large v3 Turbo (1.5 GB)

Close to large-model accuracy at roughly half the size of Large v1, v2 and v3 (2.9 GB each). The best choice when your machine can keep up.

Read about Models

Tiny, Small or Medium in between

Tiny (75 MB) is the fastest and least accurate, Small (466 MB) is noticeably better than Base, and Medium (1.5 GB) is highly accurate but slower.

Read about Models

English-only variants

Tiny, Base, Small and Medium also come in English-only versions, which are more accurate than the multilingual model of the same size if you only dictate in English.

Read about Models

A different model per hotkey

Profiles let you keep a fast model on one shortcut and an accurate one on another, each with its own language setting.

Read about Profiles

What to know first

An honest look at what local Whisper in OiPer does and does not do.

Live dictation, not file transcription

OiPer transcribes what you say while you hold the hotkey. It does not import audio or video files.

The dictionary does not apply to local Whisper

Custom words are used in cloud and hosted modes only. For a name that local Whisper keeps getting wrong, add a snippet that rewrites the misspelling, or use a larger model.

Read about Dictionary

Speed depends on your hardware

If dictation feels slow, try a smaller model or set Whisper Hardware to GPU when your machine has a usable one.

Read about Troubleshooting

Questions, answered

What people ask before they start dictating with OiPer.

Is there a Whisper desktop app for Windows?

Yes. OiPer runs Whisper locally on Windows, as well as on macOS and Linux, and types the transcript into any app. There is no Python or command line involved.

Can I run Whisper locally without Python?

Yes. OiPer installs like a normal desktop app and downloads Whisper models from inside the app, so you never set up Python, pip or a terminal.

Which Whisper model is best for dictation?

Start with Base, which balances speed and accuracy. If your computer can keep up, Large v3 Turbo gives close to large-model accuracy at 1.5 GB, about half the size of Large v3.

Does Whisper use my GPU in OiPer?

Yes, by default. The Whisper Hardware setting is Auto, which uses the GPU when one is usable and falls back to the CPU. You can also force CPU or GPU.

Should I use an English-only Whisper model?

If you only ever dictate in English, yes. The English-only versions of Tiny, Base, Small and Medium are more accurate than the multilingual model of the same size.

Does local Whisper work offline?

Yes. After a model is downloaded, transcription runs entirely on your computer and keeps working without an internet connection.

Can OiPer transcribe audio files?

No. OiPer is built for live dictation: you hold a hotkey, speak, and the text is typed into the app you are using.

Is the OiPer Whisper app free?

Local Whisper transcription is free and unlimited, and no account is required. Paid plans add hosted transcription and formatting models.

Ready to talk faster?

Download OiPer and start transcribing privately in under a minute. No account required.

Download OiPerDownload for WindowsDownload for macOSDownload for LinuxDownload for LinuxDownload for Linux

Product

  • Features
  • Performance
  • Privacy
  • Languages
  • Pricing
  • FAQ
  • Speech to text online

Guides

  • Linux
  • Windows
  • Mac
  • Whisper Desktop
  • Offline
  • Free Dictation
  • Your Own API Key
  • Wayland
  • Ubuntu

Compare

  • All comparisons
  • vs Wispr Flow
  • vs Superwhisper
  • vs MacWhisper
  • vs VoiceInk
  • vs Aqua Voice
  • vs Handy
  • vs Dragon
  • vs Windows Voice Typing
  • vs Apple Dictation

Use cases

  • Coding
  • Claude Code
  • Accessibility

Resources

  • Documentation
  • Resources
  • Changelog
  • GitHub
  • Support
  • Privacy
  • Terms
  • Refunds
Private speech transcription.

© OiPer