@mush42@hachyderm.io
@mush42@hachyderm.io avatar

mush42

@mush42@hachyderm.io

Developer of Bookworm and Sonata Speech Engine.
A proud programmer for the better part of my life.
Other interests include: machine learning, accessibility, entrepreneurship, books and music...

This profile is from a federated server and may be incomplete. Browse more on the original instance.

krys, to random
@krys@hachyderm.io avatar

If your profile boasts this flag 🇵🇸 without this 🇮🇱 next to it - as an anarchist I use neither - you are not advocating for “peace”.

You are advocating for the eradication of a democratic state that - like all democracies - is under constant and recently rather successful attacks by extremist fucks, but still has a lot of good going for going for it, and replacing it by yet another caliphate state run by a barbaric terror cult.

Am Yisrael chai.
Free Gaza from Hamas.

mush42,
@mush42@hachyderm.io avatar

@krys
Democratic state?!
Seriously?!
Israel is being ruled by a corrupted shithead, supported by a cult of extremist religious maniacs.

Tusky, to random
@Tusky@mastodon.social avatar

‼️ Big annoucement!!!

✅ Tusky 25.0 is officialy out, in F-Droid and Play Store 🎉 🎉 🎉

These are the main updates to this new version:

➡️ Support Mastodon translation API
➡️ Show post language
➡️ Improved screen transitions
➡️ Filter settings now moved to account preferences
➡️ Post stats now have a stable position
➡️ A lot off under-the-hood stability & performance improvements.

❗ Also, 25.0 app is smaller in size despite having more features.

mush42,
@mush42@hachyderm.io avatar

@Tusky
I love Tusky. Well designed app, great UX.

mush42, to rust
@mush42@hachyderm.io avatar

👋 Career change alert!

Looking to pivot into tech & leverage my 10+ years of programming experience

🐍 Python
🦀 Rust
</> Web Development
🌐 CMS: WordPress & Wagtail
✨ Machine Learning: Torch & Tensorflow

My passion for code shines through my open-source projects! Check them out:
https://github.com/mush42
https://github.com/blindpandas

Tusky, (edited ) to random
@Tusky@mastodon.social avatar

It has been nearly a week since we launched the Tusky 25.0 Beta. So far, we have encountered 3 issues in it:

➡️ crash when adding too many accounts

➡️ scheduling a status creates way too many

➡️ crash when upgrading from 24 with an Akkoma account

All of them are going to be solved soon. But we want to know: Have you encountered any other problem with the beta? Please tell us!

mush42,
@mush42@hachyderm.io avatar

@Tusky
When moving the tabs from top to bottom, the account button is no longer labled for TalkBack.

mush42, to ai
@mush42@hachyderm.io avatar

Just a random thought.
As Generative AI is being used for creating a lot of content these days, what happens when the next generation of AI models are trained using that content.
When AI models are trained on AI generated content, we'll officially enter the 8th circle of hell

mush42, to random
@mush42@hachyderm.io avatar

The second beta of sonata-for-NVDA v3.0 is out:
https://github.com/mush42/sonata-nvda/releases/tag/v3.0-beta.2

mush42, to random
@mush42@hachyderm.io avatar

New beta release of Sonata-for-NVDA, formaly known as Piper-for-NVDA
NVDA 2024.1 compatibility
Support for fast variants for Piper voices. These fast variants improves responsiveness significantly because they use streaming synthesis
Improvements to responsiveness and speed across the board
Release page:
https://github.com/mush42/sonata-nvda/releases/tag/v3.0-beta.1
Direct download link:
https://github.com/mush42/sonata-nvda/releases/download/v3.0-beta.1/sonata_neural_voices-3.0-beta.nvda-addon

mush42,
@mush42@hachyderm.io avatar

Changed the name to Sonata since we plan to support additional TTS models besides Piper in the future.

mush42,
@mush42@hachyderm.io avatar

Important notice!!!
After installing this version, you will lose all of your installed voices. Please use the voice manager to re-install the voices again.

mush42,
@mush42@hachyderm.io avatar

@ppatel @twynn
Which voices?
Custom voices or the ones downloaded from the voice manager?
I'd appreciate it if you can provide NVDA logs.

mush42,
@mush42@hachyderm.io avatar

@ppatel
It seams like the server is not running.
Are you running NVDA on a 32-bit/ARM-64 machine. Sonata only works on 64-bit versions of Windows.
Otherwise, check if the server generated any logs in the following file path:
[NVDA config directory]\sonata\logs\sonata-grpc.log
If not, try running the following binary from a cmd window and report the output:
[NVDA config directory]\addons\sonata_neural_voices\synthDrivers\sonata_neural_voices\bin\sonata-grpc.exe

mush42,
@mush42@hachyderm.io avatar

@ppatel
Maybe send me NVDA log to diagnose why the TTS server isn't running.

accessibleandroid, to android
@accessibleandroid@mastodon.social avatar

New article on Accessible Android: Sorry Google, but Geo-Restricting TalkBack’s Beta Testing Is Not the Right Decision https://accessibleandroid.com/sorry-google-but-geo-restricting-talkbacks-beta-testing-is-not-the-right-decision/

mush42,
@mush42@hachyderm.io avatar

@accessibleandroid I hate that. I cannot see any reason for it.
TalkBack should not be a product as such, it should be a system service.

mush42, to random
@mush42@hachyderm.io avatar

I think an open-source auto tagger for PDFs is very possible.
It will make it easier to convert PDFs to highly structured HTML documents.

Anyone interested in tackling this challenge with me?

Adobe already took the lead:
https://news.adobe.com/news/news-details/2023/Media-Alert-Adobe-Scales-PDF-Accessibility-With-Adobe-Sensei-AI/default.aspx

mush42,
@mush42@hachyderm.io avatar

@scruss
For this project, I'd forgo parsing the PDF stream, and extract symantic structure using a visual rendition. Then I'd use this symantic metadata to parse the PDF stream and extract text.

devinprater, to windows

So, who knew you could open a new Word document on Windows with Control + Alt + Windows + Shift + W? Why in the world? Ugh there just needs to be a "show me all the stupid Windows hotkeys that no one knows about" command.

mush42,
@mush42@hachyderm.io avatar

@devinprater
I managed to do it with one hand after the third try 😄

fasterthanlime, to random
@fasterthanlime@hachyderm.io avatar

if you hit "Ctrl+K" in Powerpoint, to insert a hyperlink, you get this delightful mix of 90s and 2010s design that I assume everyone at Microsoft is afraid to touch

mush42,
@mush42@hachyderm.io avatar

@fasterthanlime appreciate the alt text

x0, to random
@x0@dragonscave.space avatar

I am now also hosting x64 builds of Bookworm (the accessible ebook reading app) for Windows, again from the tip of the source tree. These are done since the AppVeyor artifacts expire and can be hard to find. When checking back for updates ensure you pay attention to the modification date on the file, not the version number, as the version number hasn't changed in a long time. https://x0box.xyz/uploads/bookworm/

mush42,
@mush42@hachyderm.io avatar

@x0
I wish I finish up Piper-for-NVDA faster to get back to Bookworm.
I've got many new ideas for Bookworm begging for implementation. Most of them are shooting for the moon. But hey, There's nothing impossible under the sun.
For example, I've got an idea that's nagging me for weeks: a model to automatically tag untagged PDFs. The details are already finalized in my head, but time, time, time.

jcsteh, to random

OMG. I swear I'm going to ask NV Access to back the WASAPI code out of NVDA altogether soon.

mush42,
@mush42@hachyderm.io avatar

@jcsteh
I don't want to add insult to injury here.
But WASAPI is very buggy when feeding wave samples in very small chunks. The ugliest thing is the clicking sound at the start of each wave chunk.
I'm working on piper-for-NVDA, and I've recently implemented streaming output, and this is the main show stopper.

mush42,
@mush42@hachyderm.io avatar

@jcsteh
In the range of 150-200 ms of audio per chunk.

mush42, to random
@mush42@hachyderm.io avatar

https://www.bbc.co.uk/news/newsbeat-44124396

mush42, to rust
@mush42@hachyderm.io avatar

As a visually impaired dev, my favorite sentence to hear is this:
Finished dev [unoptimized + debuginfo] target(s) in x.ys

mush42,
@mush42@hachyderm.io avatar

@ekuber
For me, Rust is one of the most pleasant languages to work with.
I like the fully integrated DX, the extremely helpful compiler, and the commandline first tooling.
IMO the formatting of compiler error messages involving complex type signitures needs to be improved. Since some types are elided I usually need to go through the message several times, and go back to the source to know what has been elided. This is because I cannot do side-by-side viewing.

Brynify, to random
@Brynify@dragonscave.space avatar

Flies are so stupid. If you land on something and get a violent protest, you would think not to land there again. But oh no no no. Not flies.

mush42,
@mush42@hachyderm.io avatar

@Brynify
You remind me of an episode of Breaking Bad where a fly is the plot of the whole episode.

mush42, to random
@mush42@hachyderm.io avatar

A new version of Piper for NVDA is out.

What's new in v2.0

• This version introduces a separate process for the TTS. This significantly increases responsiveness
• Installing voices is now easier using the integrated voice manager, which allows you to preview and download available voices
• The TTS no longer introduces unnatural pauses at the end of the line during say all
• and many other improvements and bug fixes

Detailed release notes and download link:
https://github.com/mush42/piper-nvda/releases/tag/v2.0-beta2

mush42,
@mush42@hachyderm.io avatar

@ppatel
Does your laptop has a non-Intel processor?

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • cisconetworking
  • tsrsr
  • DreamBathrooms
  • thenastyranch
  • magazineikmin
  • everett
  • hgfsjryuu7
  • Youngstown
  • slotface
  • khanakhh
  • rosin
  • ngwrru68w68
  • kavyap
  • PowerRangers
  • Leos
  • GTA5RPClips
  • modclub
  • vwfavf
  • osvaldo12
  • InstantRegret
  • ethstaker
  • Durango
  • mdbf
  • tester
  • normalnudes
  • cubers
  • tacticalgear
  • anitta
  • All magazines
    •