This is probably too niche to be worth a formal product request, so I wanted to see what kind of feedback others had on this idea first. I was thinking it’d be nice to be able to have audio wish lists. Like, there’s stuff I’d like to listen to to practice my listening, but wouldn’t be as interested in just plain reading it, and it feels wrong to add it to my normal wish list. I suppose I could add it to my wish list, then add it to a custom audio wishlist list, but then it’s sitting in my normal wish list.
Maybe this could be a consideration if some sort of dedicated audio checkbox is added for a book? Yes? No? Am I thinking too hard about this?
I think more broadly “custom wishlists” that don’t contribute to the default one would cover more use cases? I’m sure people have other use cases for it, I think some shopping sites have multiple custom lists.
I’m imagining an Audio/Listening section in the future for podcasts, radio dramas, etc. I think it would be useful to include audiobooks too, for those that don’t read the (e)book.
I think there might be a trello card for linking related media (for example, if a LN series has manga and anime adaptions), which would work to bridge between the books and audiobooks.
I think a separate listening section would be useful for tracking, to get a more complete picture of the media being consumed. Maybe it’s already been suggested, I haven’t looked.
I’ve graded quite a few books that I only listened to the audio of, I’m personally not super fussed on implementation but whether or not (and how) to pluck out those entries is something to keep in mind I suppose.
Also is Discourse updating? The links in your comment keep flashing between formatted and unformatted while I look at them
I know people would love the audiobook.jp links & filters to be autofilled out by my link collecting process… but unfortunately the audiobook.jp pages do not have any ISBN information, which makes it practically impossible for me to determine if something is a match.
However, I think there are two ways we could get it working, but I’d need help:
Some website which lists audiobook.jp links along with the ISBN or other links (like honto / amazon… anything) that I can reconcile with.
OR
Someone can find / create a simply python image matching process, which could match two images … like this and this. I would like a simple function which I input two images and receive a number on how much of a match it is.
If anyone has solutions for either of those, let me know! For the image matcher, I’m not really interested in spending a lot of time there… so if it requires training models or something it’s a no go.
Edit: And if we get no bites… that’s ok too. Eventually I’ll be tinkering with machine learning and at that point roll out an image matcher no problem. Would be useful for a lot of book providers.
You would think, but imo Amazon titles are a disaster, along with authors. Pair that with mixture of edition/formats in the title (see 魔女の宅急便)… even volume numbering … makes things more difficult.
In general, would love an image matcher, as that’s how I double check results too. But we’ll see.
We could try a text matcher but i’d think the results would be quite poor without a bit of work.
So I’m a data scientist, and while I haven’t done a lot of work outside of toy problems with image processing, I’d be super happy to work on this or future machine learning stuff for the site.
In theory the image matching piece is pretty squarely in my wheelhouse if I could find a way to do it without model training. I can poke around online and see if there’s an off the shelf solution that’d be quick to implement.
In theory that’s what we’d want, but I think Google just straight up indexes all of the images on the internet or something crazy like that so they can serve you the reverse search in less than a second. It’s a great solution when you have basically infinite storage and about 10% of the world’s total compute that you can use as long as search isn’t getting hit hard. For Natively I don’t think it’d be super practical, but it’s a good idea and probably worth checking out how they actually solve the problem.