Using the chat GPT app and BSI, I can type a prompt and send it and, a moment or two later voiceover reads the responce. I don't have to drop out of BSI or anything like that.
Now, with gemini, I'm finding that once I've sent the prompt voiceover says, gemini has sent a reply, but that's it. Also, focus has been moved out of the text input field for BSI.
I'm keen to give gemini a shot, but this is a bit of a non-starter for me. It really slows things down in a rapid back and forth. Any ideas?
Comments
My observations
I suspect the app is responding in the format with which you communicate with it, since when I use voice input, it does exactly what you want. The voice input button is always in a predictable position. I appreciate that this will not work for everyone. I have not seen a setting that would change this though.
Yeah, the live voice thing…
Yeah, the live voice thing works fine, maybe that's the better way of working it for gemini. I just prefer typing and getting responses read out automatically.
Gemini doesn't reply
Yeah, this is one of the reasons I don’t really vibe with Gemini all that much. I’ve never used Braille Screen Input before. I’ve tried a few times, but for some reason it just never stuck. Anyway, I digress.
When it comes to Gemini, one of the biggest problems I have—especially on the website—is how the responses are displayed and read out. I think this happens with the app too, although I haven’t used the app in a while so I can’t say that 100%. I remember I had a free month of Gemini set up at one point but I barely used it. Still, the issue was always kind of the same. Like, when I send a message (and to be clear, I’m not using Braille screen input—this is with the regular keyboard or dictation), it says “Gemini replied,” but then the response is super choppy. On the website, the way it displays the response makes it hard to read. Every word—or sometimes parts of a sentence—end up on different lines, and that throws off the flow of reading with a screen reader. It’s not smooth at all.
Now on the app, things are a little better. The lines are more coherent, and the structure flows more naturally. But even there, it still takes a while to load. It’s not like ChatGPT where you get this clean, consistent response almost instantly. With Gemini, the info’s there but it’s still generating while I’m trying to read it. It’ll say “Gemini replied,” but then I only get a few lines of text, and the rest of the response is still popping in word by word. It’s frustrating.
So if this is the experience others are having too, I don’t think the input method is the problem. I think it’s Gemini’s interface that isn’t designed as well as it could be, especially for screen reader users.
Funny Enough...
What y'all want with Gemini and reading out answers automaticaly, is a thing!
Though sadly only via the Google Messages Android app.
Has Google not made a Gemini app for i-Message yet?
Uh... No. It does sound…
Uh... No.
It does sound like, in the rush to be first, all of these AI developers are letting accessibility slide. I had a play with the new LLM Notes app from google and, as soon as you go into it on IOS it downgrades audio to be listening, I think it turns off one of the speakers, or something like that to avoid spill. There are also some badly labelled, if at all, buttons.
At least, when apple ever does come up with anything AI based, we will know it will work for us out the box.