epicenter icon indicating copy to clipboard operation
epicenter copied to clipboard

Enable some form of "AutomationProperties" on Windows to distinguish Mic ON from OFF state (other than visual)

Open doxgt opened this issue 1 year ago • 1 comments
trafficstars

First of all, I also back DavidGP's request on feature #67.

However, if it is deemed not conducive to implement #67, for the sake of accessibility, please consider enabling some ways to differentiate the mic ON state from the mic OFF state.

On Windows, this could be accomplished by adding different "HelpText" or Name properties for the button icons, or using "AutomationID" perhaps? This would pave a foundation for better accessibility through the likes of UIAutomation.

Many thanks in advance. This is now a core app for my daily work and personal use. A couple of customizations were essential, however:

  1. Simple post-processing for "hybrid punctuation" and translating formatting verbatim such as "new paragraph" to actual line breaks. This has markedly enhanced output readability and usability.

  2. Using AutoHotkey's "KeyWait" to implement a trivial PTT routine. This has greatly improved overall workflow. It also cuts out silence and gives opportunity to compose. Whisper may be more prone to hallucination when fed silence.

P.S., I have only used the desktop app from the HamzaETTH fork so far - thanks to DavdGP for spreading the word. That fork's minimizing-to-tray feature is great. With above customizations, for general prose dictation, Whispering is basically now "Dragon-like" for me. In fact, it does resemble a little bit of Dragon Medical One or Dragon Professional Anywhere in style and ease of use (not to be compared with desktop Dragon, with which "ease of use" generally does not go in the same sentence). The Whisper engine, moreover, trounces the medical version of Dragon on generic prose accuracy by a humongous margin.

Dragon still rules the roost for "command and control" on Windows - for average office workers at least (ambitious programmers have their own voice-driven tools, in the current "best-of-breed" of which the Whisper engine apparently also plays an essential role).

This app is a great start. It is the first and only thing so far that made whisper accessible to me personally. Thanks for the great work!

doxgt avatar May 06 '24 17:05 doxgt

No longer important to me. It would still be a "nice-to-have" feature.

Using AHK, cURL and phiola, I have been able to write a Whispering equivalent - on Windows only of course.

doxgt avatar May 19 '24 18:05 doxgt