Hi Harri,

A few notes:

1. Self voicing on iOS is not interrupting, so that appears to still need work.

2. A 2-finger swipe left when there's no text does not announce that you're at the beginning. You get no feedback whatsoever. This behavior is inconsistent with the 1-finger swipe left behavior.

3. The keyboard clicks are not the traditional iPhone keyboard click sounds. With the old MBraille, you could choose between two versions of clicks, the first being identical to the standard iPhone clicks. It would be great to have that option again.

4. The typing and associated clicks are nowhere near as responsive as the old MBraille. I type extremely fast and the new MBraille is noticeably more sluggish than the old MBraille. I just did a side-by-side comparison.

5. Is there a reason why the new MBraille always self-voices, versus integrating with VoiceOver like the old MBraille did? This could be one reason for the sluggishness.

6. I have a lot of other comments and what appear to be bugs or things that the old MBraille does better, but I'm reserving most comments and judgment until the voice can be interrupted, as the that makes things feel extremely different and awkward, so I will test everything once that's done, and I can do a much more equivalent side-by-side test.

Thanks as always!



On Jan 29, 2026, at 2:45 AM, Harri Pasanen via MBraille-Test <mbraille-test@lists.mpaja.com> wrote:


Hi all,

Release time again.   Slowly converging on MBraille 1.0, which will be a public store release and named MBraille 26.   There are still some rough edges and missing features, but I hope most of the heavy lifting is done.

Now this release is something I'm quite excited about and looking forward to hearing your thoughts.

Changelog entry below.  (Note that changelog is also a command in mbraille, where you can browse the history of changes for each release)

# Version 0.9.0

This is a rather big release with fixes and new features.


## New features

* yandex search, works like google search
* option in settings to hear clicks for char entry
* AI chat.  Currently uses Gemini.

Swipe up with 3 fingers and type ai followed by your prompt.
Example

  ai write me a haiku about fish

The answer will be inserted into your text at current cursor location.

Without arguments gemini will use your current text as prompt and append it's response to the text.  You can then continue the conversation just adding a new prompt in the end, holding dot 4 and swiping up to get your last command, which was gemini, and swiping down.

This is a very powerful interaction model, better than the chat windows available in browsers, which require you to copy paste the answers.

Important: the functionality has privacy implications, the test version uses free Gemini chat, where Google may use your data for training.  For the final production version, I will switch to paid tier, where your data is not used for training according to Google Terms of Service.

I'm wondering about announcements, now it says 'Response inserted' when you get an answer.  Short answers could be just spoken directly... maybe a preference setting for this is needed.

Anyway, let me know what you think.


## Fixes

* lang command now also changes the spelling language
* Android build in 0.8.0 had issues with liblouis, so Slovak, Danish, French at least were affected
* Number entry in French fixed
* voice feedback on number entry fixes for all supported languages
* fixed argument passing to Google, DuckDuckGo and StartPage searches.
* fixed sms command + phone number
* improved some announcements for Android.
* self voicing should now be interruptable, similar to VoiceOver.



Best regards,
Harri
_______________________________________________
MBraille-Test mailing list -- mbraille-test@lists.mpaja.com
To unsubscribe send an email to mbraille-test-leave@lists.mpaja.com