Jexxa Keeps Mac Dictation on Your Laptop for Speed and Privacy
Jexxa is Mac dictation software that transcribes speech on the device itself, with the model stored on the user's disk. The current release is version 0.1.10 as of September 15, 2026 jexxa.org.
Compatibility and privacy
Jexxa requires macOS 14 or later and Apple silicon. No Intel build is listed. Supporting one chip family simplifies what the software must run well on, namely Apple's Neural Engine for AI tasks and its unified memory design, and it avoids the performance spread of older Macs.
The website states that no audio or transcript is uploaded from the Mac, and that the software works offline without Wi-Fi. In practical terms, speech is decoded on the laptop rather than sent to a server. That removes network delay from the response time and keeps sensitive speech from traveling over the internet or sitting in remote logs.
Speed and voice editing
Jexxa's website states that most dictations are typed within a fifth of a second of releasing the dictation key. It also shows a live preview of the transcription while the user speaks. The preview lets the speaker spot errors early, while the fast final write-out keeps dictation usable inline in editors, terminals, browsers, and other text fields without breaking typing flow.
Editing can also be done by voice. The commands "JX minus one," "JX minus two," and "JX clear" delete the last line, the last two lines, or all dictation. Because these are fixed short phrases, they are easier for the system to hear correctly than open-ended instructions, and they let the user roll back misrecognitions without reaching for selection keys.
Learning, size and price
Jexxa learns corrected names and jargon after a user fix, and that learning can be undone. This addresses a common weak spot in dictation. Proper nouns, product names, API identifiers, and domain terms are uncommon vocabulary that a base model will often miss. Learning from a correction closes that gap for repeat use, while an undo path gives recourse if a correction keeps producing the same error.
It is offered in two downloads to fit different memory budgets. Full JEXXA is a 3.1 GB download for Macs with 16 GB or more memory, and JEXXA Small is a 2.0 GB download for 8 GB Macs. That choice matters in practice because the transcription model shares memory with IDEs, browsers, containers, and other resident programs.
Pricing is flat at $8 per month plus tax, with unlimited dictation and no per-minute billing.
What it means in daily use
The broader tradeoff here is spending local disk space and memory to get behavior you can count on. Cloud dictation saves local resources but ties everyday input to the internet connection, server queues, and data-handling policy. A 2.0 to 3.1 GB local package flips that around. You spend disk and working memory once, then get steady offline use and commit in under a second.
In my view, that is a reasonable exchange for professionals who dictate in source code discussions, customer notes, medical or legal drafting, or any setting where transcripts are sensitive and Wi-Fi is not guaranteed.
The broader context for evaluation is how the learning layer holds up over time. Short-term adaptation to names and jargon is useful. Long-term value depends on whether learned terms stay scoped correctly, survive updates, and stay easy to inspect and remove. Jexxa's inclusion of an undo control suggests attention to that problem, and hands-on testing with dense technical vocabulary would show how well it holds up.
