The OCR bounty’s next milestone has been reached! Previously, it was finally possible to recognize patterns of text by the OCR/translation APIs. Now, text to speech has been achieved!
NOTE: This is a work in progress/development, and is not reflective yet of the actual finished product.
In this video, you see us running RetroArch with an old school SNES RPG, Soul Blazer. Right now, this is a very experimental implementation that is far from finished and not really ready for general gameplay, but this will be further developed from this point on.
Anyway, in this video you see us doing the following:
* Whenever we pause RetroArch, the OCR process starts. It will first attempt to recognize text onscreen.
* Once it has recognized text, you will see outlines onscreen surrounding the recognized text.
* Then it passes this off to the Translate Text To Speech engine. Once done, it generates a log message about it (you can see this in the video).
* From this point on, you can unpause RetroArch and you can actually hear the voice over.
NOTE: Just to emphasize, this is NOT how it will work like in the actual finished product. This is a purely ‘WIP developer’ video we’re showing.