OpenDictate
OpenDictate is a free, open-source dictation app for Linux, macOS and Windows under MIT. It records with a hotkey, transcribes on the device with a choice of speech models (Whisper, FastConformer, Parakeet, Zipformer), and pastes the text at the cursor. One caveat: Windows support is experimental, and the live-caption model is slow on a 4-core CPU without a GPU.
What does OpenDictate do?
OpenDictate describes itself as "Free, open-source, local-first AI voice dictation for your desktop." Press a hotkey, speak, release, and the text is pasted into whatever field is active. All transcription happens on the device after a one-time model download.
Voice actions on top of the transcript include text formatting and editing commands, which the README lists as part of its workflow automation. A Models Hub inside the app lists speech engines and lets the user pick which to download. The engines are Whisper in Tiny, Base, Small and Medium sizes, Parakeet variants from NVIDIA, FastConformer CTC, and Zipformer EN 20M for live captions.
Key facts
- Licence: MIT, copyright Muhammad Waleed.
- Platforms: Linux as AppImage, .deb or .rpm; macOS as a Universal disk image for Apple Silicon and Intel; Windows as .exe and .msi, marked experimental in the README.
- Speech engines: Whisper Tiny, Base, Small, Medium; NVIDIA Parakeet variants; FastConformer CTC; Zipformer EN 20M for live captions.
- Model management: a built-in Models Hub lists engines and manages downloads.
- Processing: the README says speech is processed "100% locally on your hardware" with zero telemetry.
- Hardware: scales from a 4-core laptop CPU to a workstation; CUDA on NVIDIA and CoreML on Apple Silicon speed up transcription; no GPU is required for the base engines.
- Known limit: Parakeet Unified 0.6B streaming is "too slow for live use on a 4-core CPU" without GPU acceleration, per the README.
- Install: download the installer for the platform from the GitHub Releases page.
What does it replace, and where does it fall short?
OpenDictate replaces Superwhisper and Wispr Flow for users who want a local push-to-talk dictation app on Linux, macOS or Windows, without a subscription and without audio leaving the device.
Where it falls short: Windows support is experimental, and the README names it as such. On a 4-core CPU with no GPU, the live-caption model Parakeet Unified 0.6B is slower than real time, per the README. Whisper Medium needs about 1.9 GB of disk and substantial memory. The README names no measured accuracy or latency figure; benchmarks are left to each model's own card.
How does OpenDictate run?
Download the installer for the platform from the GitHub Releases page. Linux has AppImage, .deb and .rpm packages; macOS has a Universal disk image for Apple Silicon and Intel; Windows has .exe and .msi, flagged experimental.
On first launch, open the Models Hub and pick a speech engine. The app downloads the weights and keeps them on disk. Hold the dictation hotkey, speak, release, and the text is pasted. CUDA on NVIDIA and CoreML on Apple Silicon speed transcription up when present, and the README says neither is required for the base Whisper engines.
Who is OpenDictate for?
A Linux, macOS or Windows user who wants dictation that works offline after the first download, with a choice of speech engines inside one app. A user who wants a single engine and a smaller app should look at Handy; a Linux user who prefers whisper.cpp directly can look at Dicty on macOS and Windows. Live captions on CPU-only hardware are beyond this and most apps on this list.
What limits does the README state?
- Windows support is experimental; the main targets are Linux and macOS.
- Parakeet Unified 0.6B streaming is too slow for live use on a 4-core CPU without GPU acceleration.
- Whisper Medium requires about 1.9 GB of disk and significant RAM.
- CUDA on NVIDIA and CoreML on Apple Silicon are supported for acceleration; neither is required.
- The README names no measured accuracy or latency figure for OpenDictate; model-level figures live on each model's card.
Questions people ask
Is OpenDictate open source?
Yes. The LICENSE file at the repository root is MIT, an OSI-approved licence, copyright Muhammad Waleed. The speech models it downloads carry their own licences on their model cards: OpenAI's Whisper is MIT; NVIDIA's Parakeet and FastConformer are released under their own terms by NVIDIA.
Does OpenDictate work offline?
Yes. The README says processing is "100% locally on your hardware" and lists "zero cloud" as a feature. After a one-time weights download from the Models Hub, the hotkey works with no internet connection and audio does not leave the device.
Can OpenDictate be used commercially?
Yes for the application code, which is MIT. The speech models have their own licences on their model cards: Whisper is MIT, so commercial use is allowed; NVIDIA's models have their own non-commercial or commercial terms, so a commercial user should read the model card before shipping. OpenDictate itself adds no extra restriction.
What hardware does OpenDictate need?
For the Whisper Tiny and Base engines on a text-style transcription, a 4-core laptop CPU is enough. Medium needs about 1.9 GB of disk and more RAM. Live captions using Parakeet Unified 0.6B are slower than real time on a 4-core CPU without a GPU, and the README names CUDA on NVIDIA and CoreML on Apple Silicon as the two acceleration paths.
How does OpenDictate compare with Superwhisper and Wispr Flow?
Superwhisper is a paid Mac app with its own cleanup prompts; Wispr Flow is a paid cross-platform tool. OpenDictate does the push-to-talk and paste on Linux, macOS and Windows with a free app, and ships a Models Hub so the user chooses which speech engine to run. Cleanup is not part of OpenDictate at the time it was read. The open-source alternatives to Superwhisper page sets it beside Handy, Voz and VoiceInk.
Sources
- OpenDictate README and LICENSE: github.com/Muhammad-Waleed381/OpenDictate, read 2026-11-09.
More voice dictation apps
A free, offline push-to-talk dictation app for macOS, Windows and Linux that runs Whisper or NVIDIA Parakeet on your own machine and pastes the text at your cursor, under MIT.
A GPL-3.0 macOS dictation app that runs Nemotron, Parakeet, Cohere Transcribe, Apple Speech or Whisper on the device, with command and write modes, per-app prompts, and an optional closed local enhancement model.
A dictation and meeting-notes app under MIT that runs Whisper or Parakeet locally or cloud models with your own key, detects Zoom, Teams and FaceTime calls, and labels speakers on device.
A macOS dictation app under GPL-3.0 whose README names Superwhisper and Wispr Flow as the tools it replaces, with local models, per-app modes, a personal dictionary and an AI assistant mode.
A speech-to-text app inside the Epicenter monorepo under AGPL-3.0 that records, transcribes with a provider you choose, cloud or local, optionally polishes the text, and pastes it at the cursor.
An Apache-2.0 macOS dictation app that calls Gemini 3.5 Transcribe with your own key, writes audio to disk from the first millisecond for crash safety, and inserts polished text at your cursor without an account.
Added November 9, 2026. Every claim above comes from the project's README, LICENSE or model card, read on November 9, 2026, or from the GitHub API on the date shown in the panel. Found an error? Write to reveneau@licheo.com and it is fixed in the next weekly pass. Repository: github.com/Muhammad-Waleed381/OpenDictate.
Want OpenDictate running on your own servers?
Reveneau sets it up on your hardware, connects it to your product, and builds the software around it.