Nowado

Probably ONNX.

Last time I tried it, app literally just read papers. As in parsed arxiv pdfs text2speech. It was an awful misunderstanding of the medium. Unless it was rebuilt significantly over last months, it's just bad.

There is even more fun aspect.

'Survival' for cancer tends to be defined as surviving 5 years. The earlier you catch, the more patient had left to live anyway.

You really don't want to have white text on light grey buttons.

Fun little thing otherwise!

General tendency for internet content tends to be strong separation between creators and consumers, in particular limited interest of consumers in other consumers (think twitch chat. Each message is valued very little compared to the streamer, to the point where they always read out messages they respond to). That means unless there's something nudging people to default outside of central path of audio, adding to conversation isn't part of canon content.

There could be a way for responder to signal where the content they are answering starts, with some sort of fuzzy automation in the future. I have strong doubts about the actual experience of this for the listener, but maybe that's solvable.

I meant situation, where I already consumed the whole recording, but it gets response later on.

I do not have mental model for context being logically attached to the response. Do you think about it as response+context being a valid piece of content?

Do you have some user research you could share?

I remember thinking about this exact problem (branching conversations, in particular audio), but I couldn't find a reasonable consumption pattern.

Looking at how I consume podcasts, it's a completely passive experience - I probably have something in my hands and can't talk. Choosing paths is just too much interactivity.

I figured that maybe that's just a wrong mode to look at it and people can consume the whole thing differently, not as a podcast. Ok then, I'm an obsessed power user/fan, I consume the whole thing, all branches. Given how human attention/memory works, that means returning to earlier parts of recording after listening to branch at least some of the time, multiple times experiencing 'where did we start? Let me go back a bit. Oh, that topic was the starting. Let me forward a bit now that I know it'. That's horrible, I think. You were at least more reasonable than me when thinking about it and decided to have only 1 level of branching ; )

In similar vein, what happens when comment gets added after I already listened/how do I know which parts are 'the definite experience'? Unlike previous two issues, those questions are answerable, but I'd still like to hear what you think the answers are!

Interesting, do you happen to have some quantitative results on this/additional insights/etc?

I've interpreted transformer vector similarity as 'likelihood to be followed by the same thing' which is close to word2vec's 'sum of likelihoods of all words to be replaced by the other set' (kinda), but also very different in some contexts.

++ with mere MA in cog sci and psychology. If you really wanted to get EEG to work for typing, maybe you could train someone to map thinking about specific kinds of things to the keyboard, but that would be an extremely weird experience. Eyeball is going to have the best signal about eyeball-related motor cortex we can access.

Gives me captcha loop : (

Isn't this how copilot 'just' works, except with comments? What's the advantage over copilot?

It bothers me way less when it's not a medical treatment research.

Phonetic diagrams aren't feedback though. You would explain how to do something correctly, but user wouldn't know what elements they need to change, unless they figure out on their own what they are doing right now. I know my teachers told me to do X when they heard me mispronounce something specific, but that's irl small group/1on1 experience.

Again, I'm curious if someone solved it. Theoretically this is within modern video/audio AI, but that doesn't mean we have the models or even organized data.

I was thinking about a piece of this problem earlier and I'm curious if you figured out element that I didn't:

How do you provide feedback on how someone mispronounces a phrase?