| ▲ | libraryofbabel an hour ago | |
Thanks! I have hearing aids too and, while I don't have profound hearing loss (severe one side, moderate the other), it's got to the point where I am starting to look at other solutions to supplement them. For those not aware: hearing aids are not like eyeglasses, where you can effectively have your vision restored to 20/20 with the right prescription. They help, but they don't get you up to the level of a person with ordinary hearing. This goes particularly for situations with high background noise, multiple voices, etc.: restaurants, bars, parties and other places where social life tends to happen. Your recommendation has made me think I should finally take the leap on live-caption glasses, which I've been watching for a while now. It sounds like they've finally got to the point where they're effective. A question: how do you find them in conversations with multiple people? Do the glasses help you distinguish speakers by breaking out text from different voices, or is it all just a stream of words? What if there are multiple conversations going on - do the glasses manage to "focus" captioning on the one you're closest to, or do other conversations creep in? The captions on google meet have been useful for me as well, although I do wish they were more configurable, e.g. being able to give the captioning model a list of technical phrases, internal company terms, or acronyms that are likely to come up. There also seems to be a gap still between real time transcription, which is good, and after-the-fact transcription, which is excellent. (You can also use an LLM for the latter and prompt it yourself with extra context, although that may not be the best model.) It's been great to see so much progress in this area in the past few years. I am hoping the current (over-)investment in AI at least has the side effect of improving the tech still further. | ||
| ▲ | tlar 31 minutes ago | parent [-] | |
> A question: how do you find them in conversations with multiple people? Do the glasses help you distinguish speakers by breaking out text from different voices, or is it all just a stream of words? What if there are multiple conversations going on - do the glasses manage to "focus" captioning on the one you're closest to, or do other conversations creep in? It really depends if people are talking at the same time or not. If in a close group (space matters, the mic only goes ~2-3m), one person talking at a time, then they're great. That's the setup I most often have at conferences, so it's been great there. In evenings/parties, it's a bit more chaotic, and the glasses tend to fall off for sure. Regarding distinguishing different speakers etc I haven't needed that, I use the glasses as a helper, they fill in the gaps. So it's "just a stream of words", but it has been enough to already help a ton. The directional mic does work quite well, but sometimes you have just 2 conversations happening in front of you and it's not working at all. In other setups (especially professional ones), people usually speak one at a time and it works great. | ||