Your speech model. Your vocabulary. Corrections that consider the context.

Entune is an open-source dictation app for macOS, Windows and Linux. It runs as its own app: hold a shortcut in any app, speak, and the text is pasted where you are. Choose a cloud or local speech model, and keep your recordings on your computer. Its personal dictionary learns from your dictation; a decision model chooses when a dictionary replacement actually fits the sentence, and where paragraphs and bullets belong.

A 42-second tour of Entune on a Mac:

https://github.com/user-attachments/assets/bcbc632c-c74e-441f-a802-a3b9f4fe6d5f

Install

With uv, one command installs Entune and opens it:

uv tool install entune && entune

In Windows PowerShell, use uv tool install entune; entune.

On macOS, this puts Entune in your Applications folder, preparing it for up to a minute the first time; on Windows, in the Start menu; on Linux, in your applications menu. Then it opens Entune. From then on, open it like any other app; the terminal is no longer needed. If the terminal cannot find entune, open a new one and run entune again.

Or use pip in a Python 3.12+ environment:

python -m pip install entune && entune

uv tool install "git+https://github.com/eandualem/entune.git@develop" && entune

Start dictating: the first time, Entune opens with a short introduction, then Get started.

1. Set up a speech model: in Models, add a speech provider's API key, or download a local model. The first one becomes your default. AssemblyAI is a straightforward cloud starting point; Parakeet is our local recommendation on Apple Silicon.
2. Allow permissions: on macOS, Microphone, Accessibility and Input Monitoring, each from its own button. On Windows, only the microphone. On Linux, keyboard access: one command, run once, then log out and back in.
3. Set a shortcut, then hold it in any app and speak. The text is pasted where you are, and what you had copied is put back (on Windows, copied text only); with no text field active, the text is copied instead. Or click Record in Entune's window. Your audio and transcript are saved in History.

You can start without a dictionary or decision model and add them later.

macOS permissions belong to the Entune app, which is signed on your Mac without any certificate and stays the same across versions, so they remain granted when you upgrade. The permission guide covers missing shortcuts and “1 of 3 allowed.”

The full setup and dictation have been tested on macOS and Windows; see Windows for what differs there. Linux is new: see Linux for the two system libraries it needs and how shortcuts work on X11 and Wayland.

A dictionary match should not always become a replacement

A recognizer might write “cloud” when you meant “Claude.” But replacing every “cloud” would also damage a sentence about cloud storage. Entune stores both meanings and asks a decision model which fits the surrounding words.

The decision model selects from your dictionary. It does not generate or rewrite your dictation. Entune applies the stored spelling you can inspect and edit.

72 wrong replacements become 1 with Jev, in our test

We let Entune learn a dictionary from 7.5 hours of the maintainer's own dictation (406 recordings), then tested it on the next 2.1 hours (160 recordings) that it never saw. Parakeet transcribed everything on the Mac; GPT-6 Astra learned the dictionary on a ChatGPT plan. At the 121 places where the dictionary offered another reading, the expected answer was written down before any decision model ran (2 could not be judged and are left out):

• Jev avoided 71 of the 72 wrong replacements and kept 38 of the 47 correct ones.
• OpenAI caught the most correct ones, 42, with 8 wrong.
• Laya keeps the decisions on your computer and is the fastest, a median 0.15 s per dictation; it kept 28 correct ones.

Formatting was tested on the same dictation: of 92 places that should start a paragraph or bullet, Jev placed 35 with 12 unwanted breaks, OpenAI 8 with 1, and Laya 5 with 37, turning two whole dictations into bullet lists.

This is one person's dictation judged from the text by one reviewer, not an overall transcription-accuracy claim. Read the data, method and limits, including what the learned dictionary missed and an earlier test with the same direction.

Choose the models that suit you

Entune separates three jobs, so you can choose each independently:

Speech options also include Groq, Soniox, ElevenLabs, xAI and local Whisper.cpp. Cloud services use your own provider accounts and keys; Entune does not sell inference credits.

Use your ChatGPT subscription to build the dictionary. Dictionary setup opens on OpenAI → ChatGPT subscription: choose Sign in with ChatGPT and approve Entune in your browser. Entune uses OpenAI's documented sign-in for open-source apps, so no API key is needed, and you can set a limit for Entune in ChatGPT Settings → Usage. Your plan's model access and usage limits apply. An OpenAI API key is also available. See the access details. The subscription covers dictionary generation, not speech recognition or the decision models; OpenAI's Decisions API needs an API key.

The model guide covers exact model IDs, local-engine installation, account access, and the limits of our recommendations.

Teach Entune your vocabulary

Use Dictionary → Get suggestions to learn from the selected speech model's history. Or choose Learn from audio to import recordings from another dictation app or an audio folder. Entune transcribes imported audio with your chosen speech model, then proposes entries for review.

Dictionary generation takes time. Text goes to the suggestion model in parts, one after another, each with the dictionary entries that occur in its text. Large histories take longer; in our October test, 7.5 hours of dictation (387 transcripts) took 39 minutes in 10 parts with GPT-6 Astra at high effort on a ChatGPT plan. Imported audio is transcribed first, several recordings at a time with a cloud speech model, and parts start as soon as enough text is ready. The setup lets you choose the reasoning effort, and estimates the time from what Entune has measured. This is separate from the fast decision step on each new dictation.

Review the proposed entries before applying them. You can stop a build and review completed parts, or continue from where it stopped, with other settings if a part was too slow. Dictation keeps working throughout; only separate dictionary editing waits until the proposal is applied or discarded.

Learned entries belong to their speech model: a Parakeet dictionary is not automatically an AssemblyAI dictionary. Pin entries you deliberately want to share. Later suggestions can revise learned entries; more suggestion runs do not guarantee a better dictionary.

Keep control of your recordings

• History: replay audio, copy text, inspect processing changes, and retry a recording with another speech model. Provider failures remain visible.
• Shortcuts (macOS, Windows, Linux): hold to talk or toggle hands-free recording. Cancel without pasting; usable captured audio stays available for retry.
• The pill: a small indicator in the corner shows level bars while you record, then says what happened; an error stays there with Retry. No system notifications for results.
• Local data: audio, transcripts, settings and keys stay in Entune's data folder. Export or delete them in Settings → Data & Privacy.
• Optional processing: dictionary correction, repeated-filler reduction, and paragraph/bullet formatting have separate controls, each run by the decision model you choose: Jev, OpenAI's Decisions API, or Laya on your computer.
• Fast mode: uploads the audio while you speak, so a cloud transcript arrives sooner; the Performance chart shows each model's measured wait.
• Drop audio to transcribe it: drop files anywhere on the window, such as a recording another app could not transcribe.
• Anonymous mode: the eye switch blurs transcripts for screen recordings.
• For agents and tools: a local API accepts corrections you have confirmed, and optional Langfuse tracing shows every dictionary-suggestion request.

There is no Entune account, telemetry or hosted history. Cloud speech sends audio to your chosen provider; dictionary generation sends its selected transcripts and dictionary to the chosen language-model provider. Jev sends matched context to TypeSafe, and OpenAI's Decisions API to OpenAI, even when speech recognition is local. Laya keeps that step local. See data and privacy.

Entune is for a trusted, single-user machine. Its loopback API has no authentication: other local processes can read or change data through it. Do not expose its port to a network or tunnel.

Guides and contributing

• Using Entune: permissions, shortcuts, imports, updates and troubleshooting.
• Choosing models: cloud and local setup, recommendations and generation time.
• Decision-model results: what changed, how it was measured, and limitations.
• Dictionary reference: meanings, matching, learning and the file format.
• Building Entune.app, local API, and architecture.

To work on Entune, clone the repository and run uv sync, then uv run entune. Run uv run ruff check ., uv run ruff format --check ., uv run mypy and uv run pytest before contributing. Pull requests target develop; main receives reviewed releases. See CONTRIBUTING.md.

MIT licensed.