i am exactly like all the other guys.
ideal for comfortable gameplay in relaxed locations.
i work at, but do not speak for, microsoft.
| Web | https://halcy.de/ |
| Demogroup | https://demoscene.vc/ |
| Pronouns | He/Him |
| Priv | @halcy2 |
kind of funny to see the korean yellowpress website that pretty much randomly gets thrown onto my new tab feed at work have an article about........ research that directly cites me
nb: the newspaper article, just like every newspaper article ever written by a journalist about a paper, *completely misreads the paper*. the paper (they dont link it, but here: https://arxiv.org/pdf/2603.11877 ) is a review paper about different silent speech interfaces and is as far as I can tell, solid. Also, while "silent operation of voice systems" is a common "hey here is what you could use this for" fir silent speech interfaces, the use of LLMs that the authors of that article mean is in the traditional speech recognition sense wherein the language model is the component that helps resolve acoustic (or in this case, non-acoustic speech related biosignal) ambiguities. They don't really care about the downstream system, could be just silent telephony or going into siri or whatever you like.
hey, while I doubt there are many people here that, broadly,
1) have a phd
2) work or have enough experience to review people in the area of #speech processing, #synthesis, understanding, ...
if that is or could be you, please consider signing up to be a reviewer for #interspeech . probably too late for this year outside of late-breaking / crash reviews, but it's probably going to keep being Lots of papers.
nb: while this is, of course, unpaid work, as usual, ISCA is in fact a non-profit, and all INTERSPEECH papers are open access (and fees to attend INTERSPEECH, and publish, are and have always been kept reasonable - especially for students). So, for once, this isn't like you're lining Elseviers pockets here.
