Indie game storeFree gamesFun gamesHorror games
Game developmentAssetsComics
SalesBundles
Jobs
Tags

A tricky question about AI use for accessibility/assistive purposes

A topic by Melancholy Marionette created Jan 30, 2024 Views: 160 Replies: 2
Viewing posts 1 to 2
Submitted (4 edits) (+1)

Hey hey! Sorry to make a post on this >.< It's just a bit of a weird situation and I thought I should ask about it rather than make any assumptions :3 And sorry the post is quite long! It's a bit of a hard situation to explain, haha.

Basically, I started a cosy yandere game for Winter VN Jam in December 2023, but I couldn't manage to fully finish it by the end of the jam. I just managed to finish it properly last night, but I'm unsure if I'm allowed to submit it to this jam or not because of the way I used a form of AI in it >.<

It's related to voice acting.

I cast 2 human voice actors. 1 plays the protagonist, and the other plays the masculine voice option for the yandere. No AI was involved here, and I paid them what I could afford for their work.

The problem is, I wanted to have the lines for the narrator fully voiced as well because there are 400+ lines of narration, and myself + some of the people I know are dyslexic and find it easier to listen to long lines being read out loud than to read them just in your head.

Because there are 400+ lines for the narrator though, it wouldn't have been fair to ask a voice actor to take on so much work when I wouldn't be able to afford to pay them properly for their time >.< So I recorded myself reading all the lines, and then I used speech-to-speech dubbing to change my voice (because my voice is horrible and my mic isn't great!)

I also wanted to try out voice acting myself, but I have crippling anxiety >.< so I decided to try and voice the feminine voice option of the yandere the same way by recording myself performing all the lines and using speech-to-speech to dub over them. It meant I could have a go at voice acting without actually having my horrible voice in the game, haha. Because the speech-to-speech kept my delivery of the lines, all my intonation,  pauses, breaths, etc. intact.

I'm not a fan of AI, and this is the first time I've ever used it in a project. It was more as a last resort sort of thing because if I hadn't, then the narration lines and the feminine voice version of the yandere character would all be silent/unvoiced since I wouldn't have had the time or money to get proper voice actors to fill those roles on this particular project >.<

I have been open about it in a dev post as well so that people are aware.

No worries if you would prefer I didn't submit the game still! I can just about justify my reasons for using it to myself, but I totally understand if it's not acceptable for this jam :3

I'm going to try my best to make a completely different & brand new mini-project from scratch specifically for this jam sometime anyhow <3 It will be short because I won't have much time to work on it while the jam is running, but I'm still gonna try and make something ^-^

Thanks and have a great day :D (and also thank you for making the jam :3)

Host

Hmm, I'm unsure. I can understand your reasons, and I'm tempted to make an exception, but it would probably be fairest to not allow it- the only thing I'm concerned about is what you used to do this; as you may have consented to using your voice to train the AI, but this AI may have been trained on other voices without their permission, if that makes sense.

Because of that I feel it would be best not to allow it, I'm sorry. But if anyone disagrees, please let me know, I'm willing to hear what others think about this.

Thank you for bringing this up, have a great day ^^

Submitted

No worries at all! I completely understand cos it's a bit of a weird situation and somewhat dodgy ground, haha. That's why I thought it was best to ask, so thank you for getting back to me :3 

The service I used didn't require me to consent to them using my voice in any way. It stated that they won't store any data from my uploaded recordings, but really I have no way of verifying if that's true or not >.< And even though they say their training data is all ethical for their voices, again, there's not really any way for me to truly verify it, so I can absolutely understand that.

You don't need to apologise anyhow cos it's fair enough :3

If it wasn't for the fact that I was rushing to make this during winter jam last December, I may not have had to resort to using the speech-to-speech dubbing anyhow because I would've potentially had more time to gather more funds to pay a proper narrator, but the jam was only a month long, and I hadn't actually planned to join it, so it was all a bit of a mad rush xD I'd told myself no more jams for 2023, and then December came along and I wound up joining anyways, haha.

Had the project been started for this jam, I would've spent more time on it and wouldn't have even touched the speech-to-speech stuff. I doubt I'll ever use it again either unless it's for another emergency x3

With any luck, I'll have the time to start a brand new mini-project for this jam before the jam ends! :3

Thanks again and have a happy last day of January!