Totally. I love the idea but my guts says comparing someone’s voice with stock audio (reliably and at reading-speed) is not technology we have readily available right now. I’d love for someone to proove me wrong though.
Ultimately, I think a brain can tell when it’s wrong- instantly, and for free. Maybe it makes more sense to select specific characters that you’re struggling with so that you get more iterations on those?
This reminds me of an xkcd classic which always makes me laugh…
Also a good idea. What if you tapped a button while you read? あ-tap-い-tap-う-tap-え. It’s more work, but would make that calculation trivial. Maybe each character would highlight, and tapping advances the highlight?
Otherwise, I could just time how long it took you to click “Next”, but I don’t even know if people use that feature, so it might not provide helpful information.
I’m not sure, but please let me know if you find out empirically.
I’d love that. Link to the repo is at the bottom of the page. Happy to discuss more in GitHub if you need any help, just open an issue.
Yeah, great. Let’s do it. What kind of font do you want?
Thanks for the feedback, everybody!