free & open source

Talk. It types.

Hold fn, speak naturally, release. Your words appear instantly — cleaned up, formatted, and ready.

Download for  Silicon Download for  Intel

Not sure? Apple menu → About This Mac → Chip

Download for Windows View Source

100% free · macOS 13+ / Windows 10+ · MIT Licensed

How it works

1

Hold fn

One key starts recording from your microphone.

2

Speak

Whisper transcribes your voice in real time.

3

Release

AI-polished text appears right at your cursor.

Built for speed and privacy

Lightning Transcription

Powered by OpenAI Whisper. Fast, accurate speech recognition that handles accents and jargon.

AI-Polished Output

GPT removes filler words, fixes grammar, and formats your speech into clean, readable text.

Private by Design

Audio is never stored. Your API key stays on your machine. No accounts, no telemetry.

Fully Customizable

Open source Swift. Fork it, tweak prompts, change the hotkey, add features — it's yours.

Make it yours.

CustomWispr is free and open source. Copy this prompt into any coding agent to get started:

Fork and clone https://github.com/beausterling/CustomWispr — it's a speech-to-text app for macOS (Swift) and Windows (C#/.NET). Read the README and codebase, then help me customize it.
View on GitHub Support the Project

Frequently asked questions

Yes. CustomWispr is a 100% free, open-source alternative to WisprFlow. It provides the same core functionality — hold a key, speak, and get AI-polished text — without any subscription or payment required. Just bring your own OpenAI API key.
CustomWispr offers similar speech-to-text functionality to SuperWhisper but is completely free and open source. It uses OpenAI Whisper for transcription and GPT for text cleanup, runs from your system tray or menu bar on both macOS and Windows, and requires no subscription. Because it's open source, you can customize everything.
Yes, CustomWispr is 100% free and open source under the MIT license. You only need your own OpenAI API key, which typically costs fractions of a cent per use. There are no subscriptions, no accounts, and no hidden fees.
Yes! CustomWispr is fully open source, so you can customize it to work with any API provider or local LLM. Whether you prefer Ollama, LM Studio, Anthropic, Groq, or any OpenAI-compatible endpoint, the code is yours to modify. Check out the GitHub repo for details on how to swap providers.
Unlike Apple's built-in dictation, CustomWispr uses OpenAI Whisper for far more accurate transcription — especially with accents, technical jargon, and natural speech. It also runs your text through GPT to remove filler words, fix grammar, and format output cleanly. The result is polished, ready-to-use text.
Yes. CustomWispr never stores your audio. Your API key stays on your machine. There are no accounts, no telemetry, and no data collection. Audio is sent to OpenAI's API for transcription and immediately discarded.