Speech recognition: worked really well. Suspiciously well, even for random technical content to do with gudgeon pins and cars. Knew when to capitalise some obscure words. Seperates sentences. Seems to be Whisper-Web.
Text to speech: some of the “accents” are not very good or skip words. Still though nicer than espeak. No idea where the code is for this or if it’s based off another project, no info provided.
Enhance image: I ignored the warning and started processing a small image on my small laptop anyway. One CPU core used for 5 or so minutes and 8G of ram eaten, but then nothing happens. Didn’t even start my laptop’s fan or make my OS swap. Killed/crashed?
The rest: links to offsite stuff so I didn’t test it.
Overall pretty neat, but with no readable source (except for the speech recognition) I have no easy way of verifying that this even is running 100% offline, nor learning from it, no adapting it.
Tested on my low-end laptop:
Overall pretty neat, but with no readable source (except for the speech recognition) I have no easy way of verifying that this even is running 100% offline, nor learning from it, no adapting it.
EDIT: Looks like it might be a mix of these? https://github.com/xenova/transformers.js
is this open source?
I couldn’t find any mentions of license on the page and the page source is all on one line.
From some orange-site discussion it looks like it might be a collection of these: https://github.com/xenova/transformers.js