Neither interns nor students are expensive, the latter are free and are begging to work for you. I can get them in the snap of a finger and I work in a much sma

Neither interns nor students are expensive, the latter are free and are begging to work for you. I can get them in the snap of a finger and I work in a much smaller university. For reference Berkeley got 5 billions through fund raising campaigns since the ADA was implemented, regularly gets a 100 million dollars from donors in a year, and enrolls 40000 students every year (there's your 40000 hours) who pay tuition. Also they committed to adhering to ADA for every new piece of content they produce. What I proposed is text-to-speech with a human in the loop, semi-automatic transcription. I've worked as a data scientist and machine learning engineer then researcher in the industry and in academia, on speech recognition and generation models among other things. I've labelled time-series datasets containing hundreds of thousands of items by myself. It's not rocket science, there are very efficient tools, and if you can't find one for your purpose it's a couple of weeks of work to create one. The ADA captioning requirements aren't very complicated either, you can check them out yourself it's like 5 things.