For NVDA Users: ObjPad Add-on 26.02 https://groups.io/g/tech-vi/message/11223

glad to find this. Just getting into markdown, so this will be nice for moving around in them. https://github.com/cary-rowen/markdownNavigator

@kaveinthran oooh, it worked! "Notification Done! $50 in extra usage added to your account. " that's super awesome

@Tamasg nt sure if you are awear, but just in case,
"We granted all current Claude Pro and Max users $50 in free extra usage. This credit can be used on fast mode for Opus 4.6 in Claude Code.

To use, claim the credit and toggle on extra usage on claude.ai/settings/usage. Then, run `claude update && claude` and `/fast`. Enjoy!"

@Tamasg nt sure if you are awear, but just in case,
"We granted all current Claude Pro and Max users $50 in free extra usage. This credit can be used on fast mode for Opus 4.6 in Claude Code.

To use, claim the credit and toggle on extra usage on claude.ai/settings/usage. Then, run `claude update && claude` and `/fast`. Enjoy!"

OK -releaseing an experimental NVDA add-on for SpeechBox that has CMU-look up for word-level overrides to Espeak. We align phrase-level, and if a word matches the lookup in the CMU via the frontend, we replace it from Espeak's. This should make US English sound more natural.
https://eurpod.com/synths/tgSpeechBox-2026.nvda-addon
Or just,, update via the updater in the language pack settings window last item. I'm not releasing yet because I want feedback on the reading changes here. UK English should remain untouched.

Crazy! https://www.facebook.com/?force_mobile=1 now works again. Seriously whoever you are thank you. Less tech support for me in the immediate future.

Show HN: LocalGPT – A local-first AI assistant in Rust with persistent memory

Link: https://github.com/localgpt-app/localgpt
Discussion: https://news.ycombinator.com/item?id=46930391

soundWave: An NVDA TTS recorder that records *some* synths directly without need of a virtual cable, mixer or even an audio interface at all.
This is very beta software and it may eat your speech for breakfast. I'm warning you, be sure to have your NVDA reload key on-hand.
Currently supported synths:
DecTalk, Eloquence, Orpheus, SAPI5 and Sonata.

Download: https://onj.me/nvda/soundWave.nvda-addon

Guide: porting NVDA add-ons to 64-bit NVDA (2026.1): https://nvda-addons.groups.io/g/nvda-addons/topic/guide_how_to_port_add_ons_to/117685297
TL:DR: check the add-on code, look for 64-bit dependencies if needed, test the add-ons, then publish it (hopefully to the add-on store).
Reason: add-ons are crucial part of the NVDA experience. Porting add-ons to 64-bit NVDA keeps them and NVDA relevant. The greatest thing that can happen to the community in 2026 is discovering the joys of using add-ons once again in the midst of celebrating NVDA's 20th anniversary.

@fastfinge The story of what actually happened there is a bit sad. I first translated NVDA to Hungarian in 2007, then later a dedicated group came together to maintain it from Hungary. They contacted the organization who licenses the voice for JAWS (Profivox at the time) to get it into NVDA at low cost or free. The organization, which is a blindness-specific one kind of like an "NFB" type here, said that NVDA is in direct competition with JAWS and because they are FreedomScientific's prime distributor, they will not allow ever the use of the engine in NVDA. That was that. Kind of sad when you see everyone else outside of you as "just the competition" and can't live with the idea of coexistance. It's why I don't trash other formant Klatt projects, never will,, people are putting time into them.

@fastfinge also bundling Onnxruntime (but maybe easier as a DLL module not full Python bloat?) for older NVDA. This would be smaller than your Gruut experiment: I'm looking at about 11 MB for the lightweight model, and then another 15-20 MB for onnxruntime DLLS, then anothr 20 MB for numpy. Which it also requires. So in total the add-on would become 80 MB.
We could do a mechanism where we use this G2P for English only but other G2Ps or Espeak for foreign languages.

Even though I think 1password has better accessibility, I still think over all bit warden wins because I haven't found a page yet where detection of logins didn't work automatically. With 1password old.reddit.com login wouldn't work unless I clicked the extension in the the chrome toolbar: The sub menus in the right click menu also help with BitWarden for me and make up for the weird arrowing issues with it.

Wish BitWarden would fix a few accessibility concerns. 1password deffinitly leads in accessibility, and it's slightly better unlocking than bitWarden is. I have to check autofill and only lock on reboot at the very least in order to have a good user experience with it. https://www.makeuseof.com/switched-from-lastpass-to-bitwarden-open-source-password-manager/

@Tamasg Are you planning to play with this? Can be a novel experiment to have crowd source pronunciation toolspowering #screenReader TTS

The Wiktionary thing genuinely excites me though. Imagine we've got 27 language packs right now. Even if only 5-6 of those languages benefit from a pronunciation dictionary, that's thousands of stress corrections per language, all from one unified pipeline, no new dependencies.

All my pr for ace-step are merged. Now you can use their main branch and enjoy all the benefits of their speedy improvement!
git clone https://github.com/ace-step/ACE-Step-1.5
uv sync
source .venv/bin/activate
python cli.py
@ZBennoui @vick21 @FreakyFwoof

A lot of people told us they wanted to try Perspective Intelligence but did not have the newest iPhone. We heard you. Version 1.4 added the Gemma 3 4B model so you can use the app on older devices too. It is a little slower right now but we are already working on making it faster in the next update. We want this to be for everyone.
https://perspectiveintelligence.app
#OnDeviceAI #iOS #AI #IndieApp #Apple #Tech #Gemma

Nice list. https://gist.github.com/Molitvan/50e3b5060ab9465b1da895155d5c0480?permalink_comment_id=5930768

"And then there's a sneaky third option — Wiktionary dumps. Wiktionary has IPA transcriptions for hundreds of thousands of words across dozens of languages, and it's CC BY-SA 3.0 (compatible with GPL v2). Someone has already parsed these into structured data — the project is called wiktextract. That could be your multilingual dictionary source without needing to find a per-language resource."
OMG Opus 4.6,, you're killing me tonight with your brilliance.