ashraymalhotra

In our start-up, we make edits to the video content (minor edits, like slight changes to the face etc) to make the content more engaging for the audience.

To understand our product market fit, we need to understand if the retention for our videos is better after our processing (we do this for advertisers, so better retention means higher returns for them). I was looking for tools which with a webcam feed can track how "attentive" the person is, but I could not find anything good.

The best resource I found was https://www.realeyesit.com but they are asking us for 17.5k USD (which is obviously unaffordable for a pre-product startup). Some other resources like https://xlabsgaze.com are highly inaccurate.

Do open source tools exist for video engagement estimation using webcam? Alternatively, are there more accessible tools for this purpose (under $100 per month).

I was talking to a friend who's cousin has been deaf and dumb since birth. They tried giving her "implants" to make her hear but the background noise made it intolerable. Given the breakthrough in Voice to text technologies recently, I took it for granted that she would be able to understand any conversation around but apparently she isn't able to make out anything people say and sometimes gets sad because of that (on quite a few occasions she can successfully lip read though).

Something even as simple as a continuous background voice to text converter, and when she wants to "talk" a text to voice converter should be much better for her than her current status (my assumption is she always has a laptop or mobile screen in front of her).

This should be as simple as integrating some Voice to text and text to voice APIs from anywhere (say Google cloud, assuming internet connectivity isn't a problem?)

This seems incredibly basic and I assume systems like this should exist! If not, I could hack it over a weekend. Could you please link me to the existing tools you know to solve this problem for folks with hearing problems?