soupslurpr @ soupslurpr @fedia.io

Posts

1

Comments

14

Joined

1 yr. ago

You can still get the apk on GitHub releases if you don't want to use Accrescent. Also you can use Obtanium to get automatic updates for many apps from different sources, including GitHub.

Have a nice day to you too, and your welcome! Also keep in mind Transcribro can be used the same way as you imply you're using Sayboard with FUTO. Just saying in case you didn't know.

Please don't spread baseless claims. The majority of the apps on Accrescent as of writing are not made by me. Only 3 out of the 12 apps are. I have already stated if you don't want to get my app from Accrescent, then you don't have to. There are apk files released on GitHub releases, which the download section of the README.md talks about.

For a comparison, the developer of Accrescent has said some of the differences in an issue for adding a comparison table (it hasn't been added yet). See https://github.com/accrescent/meta/issues/22#issuecomment-1780020882

The main selling points of Accrescent are no accounts required, unattended updates without root (on Android 12+), developers always sign their own apps, and the security properties of Accrescent which means even if Accrescent's servers gets compromised, attackers can't do something like replacing an existing app with their own malicious version without access to the signing key used to sign the repository metadata.

Also keep in mind Accrescent is still in early alpha and so it doesn't have all of the features that are planned. If you'd like to see some of the planned features, please take a look at the roadmap at https://github.com/orgs/accrescent/projects/4/views/1

Yes, please use GitHub's "Issues" tab and you'll see an entry has been made to support multiple languages.

For now, it is English only.

Have you read their FAQ at https://accrescent.app/faq ?

There! I added a screenshot of the keyboard UI at the bottom of the README.md. There isn't really anything else to screenshot other than maybe the ACTION_RECOGNIZE_SPEECH activity but it's basically the same but floating and with only the mic button.

I understand. I'm not fond of the three options you presented.

Hi, you don't need to download another app store to get Transcribro. Accrescent is not required, just recommended (because of its security properties and optional auto updates). If you read the download section, it clearly states that Transcribro is also available on GitHub releases. You could use something like Obtanium if you want to get it from GitHub releases with auto updates, or just install the apk. You don't have to use any app store 😄

Yep. Also Transcribro has this feature that makes editing easier as it adapts the punctuation https://github.com/soupslurpr/Transcribro/pull/39

Again, you can use the FUTO keyboard and Transcribro together. All you have to do is enable Transcribro's keyboard (the Start screen will guide you), and then make sure in FUTO's keyboard in Voice Input the "Disable built-in voice input" is on. After that, pressing the mic button should switch to Transcribro. If you'd like Transcribro to automatically switch back, you can toggle that on in its settings.

Screenshots would be nice I suppose, yeah. A walkthrough/instructions are sort of included in the app in the Start screen, at least enough to get stuff enabled. I think the keyboard UI is pretty easy to understand, and the buttons are large and there is no wasted space which is nice for accessibility. At this time there aren't any plans to publish it on F-Droid. However, Accrescent is the recommended way to get it as Accrescent provides automatic updates (which can be turned off if you'd prefer) and has higher security than GitHub and F-Droid. It's available on GitHub releases too.

Thanks, an F-Droid or IzzyOnDroid release isn't planned at the moment. Sayboard uses the Vosk model and because of that it can have almost real time transcription. It also has a punctuation keyboard integrated within. But Transcribro has better accuracy, punctuation, and capitalization as it uses OpenAI's Whisper model. While Transcribro isn't real time, it will compute very fast after it stops detecting speech (which takes a few seconds due to not wanting to stop early) or press the mic toggle to stop it yourself. On shorter stuff on my Pixel 7 it can take less than two seconds. There are many more differences and I'd suggest trying it out yourself to see exactly.

FUTO keyboard seems pretty cool! By default it switches to an enabled voice input keyboard when you press the microphone button, so you can actually use Transcribro with FUTO's keyboard! Better together! Of course perhaps you like their voice input or another voice input more. But it's good that they give you an option rather than being limited to only theirs! Transcribro might be a better option for you if you like how it has more buttons and some unique features compared to FUTO's. It's also around the same speed and accuracy as they both use the same underlying model. So give it a try and let me know!

Haha it used to work when I made it and tested it with Bing with access to the current page. Just tested now with Copilot and it did not work. I think the reason why is because they have been trained to be more resistant to prompt injection than before OR some copyright sensitivity training done on it, which is suggested by the chat shared above.

P.S. For proof this is the real soupslurpr (the developer of BeauTyXT) check my GitHub profile it has a link to my account on this site. https://github.com/soupslurpr