JEXXA — Dictation that never leaves your Mac

JX
JEXXA
Dictation that never leaves your Mac.

Download for Mac
3.1 GB · 16 GB or more

Download Small
2.0 GB · 8 GB Macs

macOS 14 or later · Apple silicon · version 0.1.10

Everything happens on your Mac
The model is on your disk. Hold a key, speak, and the words
appear where your cursor already is — in any app that takes text.

🔒
Nothing is uploaded
No audio, no transcript. What does leave is small and listed in the
privacy policy: your subscription check, update checks, and
diagnostics you can turn off.
⚡
Fast enough to think in
Most dictations are typed within a fifth of a second of letting go of
the key. No queue, no per-minute cost.
✏️
It learns your words
Names and jargon it gets wrong once, it gets right after you fix it.
It tells you what it learned, and you can undo it.
↩︎
Take back what you said
Say JX minus one to remove the last line, JX minus two
for two, JX clear for the lot.
👀
See it as you speak
A live preview while you talk, so you know it is hearing you.
✈️
Works with the wifi off
On a plane, on a train, on a bad hotel connection. It does not care.

Which one do I want?
Same app, same features. The difference is how much of your
memory the model needs while it is running.

16 GB or more
JEXXA
3.1 GB download ·
the more accurate weights
Download

8 GB
JEXXA Small
2.0 GB download ·
smaller weights, leaves your Mac usable
Download Small

One price
No per-minute billing, because nothing is being processed
anywhere but on your own machine.

$8 / month + tax

Unlimited dictation, anywhere you can type
Runs on your Mac — no upload, no queue
Learns the names and words you use
Your voice and text never leave the machine

Home
Privacy
Terms
hello@jexxa.org
Apple, the Apple logo and macOS are trademarks of Apple Inc.,
registered in the U.S. and other countries.

JEXXA is a dictation system designed for macOS that processes voice input directly on the user's machine, emphasizing extreme privacy and high performance. The core functionality allows users to dictate words by holding a key and speaking, with the resulting text appearing precisely where the cursor is located within any application that accepts text. A significant architectural feature of JEXXA is its commitment to local processing; no audio or transcripts are uploaded, ensuring that all sensitive data remains on the Mac, adhering to a strict privacy policy where only subscription, update, and diagnostic checks are noted as potential data transfers.

The system is engineered for exceptional speed, allowing dictation to occur within a fraction of a second of releasing the key, eliminating concepts like queuing or per-minute billing. This performance is tied to the fact that no processing occurs remotely, as the operation is handled entirely on the user's local hardware. JEXXA incorporates a learning mechanism that allows it to adapt to the user's specific vocabulary, correcting errors in names and jargon after the user provides feedback, and retaining the ability to undo learned information. Furthermore, the system provides a live preview while speaking, giving the user real-time confirmation that the system is actively hearing and processing their speech.

The operational flexibility is substantial, as the system functions independently of an active internet connection, making it useful in environments with poor connectivity, such as air travel or remote locations. For users concerned with local processing versus system resource allocation, JEXXA offers two distinct model options. The larger JEXXA model, requiring 16 gigabytes or more of memory, provides the most accurate weights for operation. Conversely, the JEXXA Small model requires only 8 gigabytes of memory and features smaller weights, making it suitable for keeping the Mac usable while still providing dictionary functionality. Regardless of the chosen model, the service operates under a single monthly fee of eight dollars plus tax, with no associated per-minute costs.

Ultimately, JEXXA integrates advanced dictation capabilities with robust security by ensuring that the user's voice and text never leave the physical machine. This focus on on-device operation, combined with sophisticated learning and control mechanisms, positions the application as a private and high-performance solution for text input on Mac systems.