Entune
Cross-platform dictation app where a decision model chooses whether a personal-dictionary replacement actually fits the sentence.
Overview
Entune is an open-source dictation app for macOS, Windows and Linux. Hold a shortcut in any app, speak, and the text is pasted where you are, using a cloud or local speech model of your choice, with recordings kept on your computer. A personal dictionary learns from your dictation, and coding agents can build and refine it over MCP.
A decision model chooses when a dictionary replacement fits the sentence and where paragraphs and bullets belong. For example, a recogniser may write cloud when you meant Claude, but replacing every cloud would damage a sentence about cloud storage, so Entune stores both meanings and asks which fits. The model selects from your dictionary and never generates or rewrites dictation. In the maintainer's test on 2.1 hours of held-out dictation, Jev made 1 wrong swap out of 72 matches that should stay and caught 38 of 47 needed fixes, compared with Perplexity, OpenAI and Laya, against a baseline that swaps every match.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which stored meaning fits this word in context: keep it or replace with the dictionary spelling? | — |
| Choice | Where should paragraphs and bullets go? | — |
How it works
- Record speech and transcribe with the chosen model
- Find personal-dictionary matches
- Decision model chooses whether each replacement fits; applies formatting choices
- Paste the result into the active app
Strengths
- Avoids blanket find-and-replace errors
- Model only selects stored spellings
- Several decision providers compared on real data
Caveats
- Evaluation is on one person's dictation
- Cloud decision models receive transcript text
Getting started
uv tool install entune && entune (or python -m pip install entune).At a glance
- Type
- Project
- Status
- beta
- Language
- Python
- Runtime
- desktop app (macOS, Windows, Linux)
- Has code
- yes
- Has tests
- yes
Use cases and integrations
How curated lists describe it
Open-source, cross-platform dictation with your choice of speech model and a personal dictionary where Jev selects corrections in context.
hellogumbo/awesome-jev
Frequently asked questions
What does the decision model do?
It chooses whether a dictionary replacement fits the sentence and where formatting goes.
Does it rewrite my dictation?
No, it only selects from your stored dictionary spellings.
How did Jev perform?
In the maintainer's test, 1 wrong swap out of 72 and 38 of 47 fixes caught.
Related projects
Liquid Glass voice typing for macOS 26 with an optional Jev fast path that types clean short sentences without waiting for LLM cleanup.
ProjectbetaSwiftMac voice assistant where one Jev call per utterance maps speech to app, volume, Spotify and system actions; an LLM only answers questions.
AppprototypePythonBrowser teleprompter that checks off talking points as you cover them, scoring each with a parallel Jev Score question.
DemoprototypeJavaScriptIDE and browser extensions to find and install agent Skills and Subagents, with an opt-in Jev check when local detection can't decide.
Browser extensionproductionTypeScriptTriage for DingTalk mail and chats where Jev sorts messages into needs-reply, updates, promo, sales or spam with attention flags.
ProjectbetaPythonLocal-first meeting transcription and conversation memory app whose live voice path runs its evaluations through Jev.
AppbetaRust