Pondering the Practicality of Voice AI

Hey yall. I was watching some HC WoW streams and almost everyone is using the Voice AI addon that vocalizes all the quest and dialogue box text. I think this is a simple (from the user perspective) way to improve the immersion of the game and also great for those who need more accessibility options.

My question for you is about how practical this would be for Intrepid to implement this into their own game (seeing as this game will not be allowing addons). Is it a matter of price? A matter of time? Maybe it would expand the file size of the game too much?

I don't know much about the new AI tools, but it seems like a somewhat simple and easy process where maybe each NPC has a couple parameters, like gender, accent, age, and then as you interact with the NPC the dialogue gets generated into an audio clip for you to listen to. Also, I assume you could license one of these tools from another company to save time and resources in generating your own AI tool.

Let me know your thoughts. Cheers.

Find more posts tagged with

Staff

Comments

Nerror

We've had a few threads about this before, and to repeat myself, I 100% think they should implement AI voice over of all possible text in the game as an option. As I see it, it's a really important accessibility feature. No expensive voice actors, and no other cost than the AI system itself and implementing it into the core functionality and UI. It doesn't have to be perfect or anything, because it's an accessibility feature. It shouldn't replace proper voice actors in their cut scenes.

Caww

keep it on the cheap (don't raise the monthly), don't drag the release time out any further and I'm good...

McShave

Caww wrote: »

keep it on the cheap (don't raise the monthly), don't drag the release time out any further and I'm good...

I would like to know how much licensing this kind of stuff would be. Someone made that add-on for WoW and people can download it for free, so why would you not do that, you know?

Veeshan

If it a cost/time viable option i think they should, it add so much more to the world i think especialy since most poeple wont read the quest but they listen to dialog

Depraved

Nerror wrote: »

We've have a few threads about this before, and to repeat myself, I 100% think they should implement AI voice over of all possible text in the game as an option. As I see it, it's a really important accessibility feature. No expensive voice actors, and no other cost than the AI system itself and implementing it into the core functionality and UI. It doesn't have to be perfect or anything, because it's an accessibility feature. It shouldn't replace proper voice actors in their cut scenes.

if the case for ai voice is accessibility, then is can use a regular screen reader to read the texts out loud. no need for ai. its also probably cheaper and faster as well.

mk1 has one and sounds horrible, and its turned on by default T_T

Noaani

The major issue with this right now is from a legal perspective.

As it stands right now, any AI work is inelegable for copyright. This hasn't gone through court as yet, but there is the possiblity that once it does, any work using any AI in it's finished product may be inelegable for copyright in it's entirety.

This is actually a real, plausable scenario that some very large (and litigious) companies including a specific mouse-run entertainment company with themeparks (not willing to name the company, due to it's ligitious nature) are wanting to see become reality, and so I wouldn't go betting against it becoming law in America at some point in the future.

If this did indeed become law, it could mean that any game shipped with AI voice as standard (as opposed to as a mod) could be stripped of any copyright status, essentially making it public domain.

While it's easy to stand by and say that you don't think this would happen, you have to ask yourself what sort of odds you are willing to bet tens of millions of dollars on.

I know it's a bet I wouldn't want to go anywhere near right now, not until it's gone through the court (ideally the Supreme Court).

Depraved

you can still copyright the assets of your game (art and sound), the name, the logo...

daveywavey

Maybe it's time for a Mega-thread.

Songcaller

Noaani wrote: »

The major issue with this right now is from a legal perspective.

As it stands right now, any AI work is inelegable for copyright. This hasn't gone through court as yet, but there is the possiblity that once it does, any work using any AI in it's finished product may be inelegable for copyright in it's entirety.

This is actually a real, plausable scenario that some very large (and litigious) companies including a specific mouse-run entertainment company with themeparks (not willing to name the company, due to it's ligitious nature) are wanting to see become reality, and so I wouldn't go betting against it becoming law in America at some point in the future.

If this did indeed become law, it could mean that any game shipped with AI voice as standard (as opposed to as a mod) could be stripped of any copyright status, essentially making it public domain.

While it's easy to stand by and say that you don't think this would happen, you have to ask yourself what sort of odds you are willing to bet tens of millions of dollars on.

I know it's a bet I wouldn't want to go anywhere near right now, not until it's gone through the court (ideally the Supreme Court).

The written format is automatically copyrighted. So, only the vocal part can't be copyrighted but the written form is copyrighted. Thus, anyone can already read written words and verbalise them (can't be copyrighted only digitized rights for digital products). So, unless the ai has produced the written word, verbalised the written word and published the written word copyright ambiguity does not apply.

Caww

McShave wrote: »

Someone made that add-on for WoW and people can download it for free, so why would you not do that, you know?

you can't beat free...

Balrog21

This has been the only time since mid Cata have I played WOW, and that was only to see how well the AI voiced quests were done. To my surprise it was an eye opener for sure, it did help with immersion 100%. It was also nice to start the dialogue and listen while you were headed to the quest area.
I think it would be a no brainer for Steven and the gang to implement this. Providing no legal issues would come about later on down the line.

Voeltz

Definitely worth pursuing. Adds another layer of depth that can't be achieved just by reading text. There are several routes they could go with it, either hire a few devs that have experience working with AI voice apps or outsource it. Each voice line can be generated in maybe 10 seconds and it's as simple as copying the text, pasting it in the module and clicking a button. It might take a few tries to tweak the settings just right for a good sounding voice line, but it's entirely worth it. Here's a company that does it for Obsidian Studios, keep in mind this is 2 years old as well.
https://m.youtube.com/watch?si=Gsf7ASZxbhoXHzmW&v=YajBa5PO1Hk&feature=youtu.be

Noaani

Neurath wrote: »

Noaani wrote: »

The major issue with this right now is from a legal perspective.

As it stands right now, any AI work is inelegable for copyright. This hasn't gone through court as yet, but there is the possiblity that once it does, any work using any AI in it's finished product may be inelegable for copyright in it's entirety.

This is actually a real, plausable scenario that some very large (and litigious) companies including a specific mouse-run entertainment company with themeparks (not willing to name the company, due to it's ligitious nature) are wanting to see become reality, and so I wouldn't go betting against it becoming law in America at some point in the future.

If this did indeed become law, it could mean that any game shipped with AI voice as standard (as opposed to as a mod) could be stripped of any copyright status, essentially making it public domain.

While it's easy to stand by and say that you don't think this would happen, you have to ask yourself what sort of odds you are willing to bet tens of millions of dollars on.

I know it's a bet I wouldn't want to go anywhere near right now, not until it's gone through the court (ideally the Supreme Court).

The written format is automatically copyrighted. So, only the vocal part can't be copyrighted but the written form is copyrighted. Thus, anyone can already read written words and verbalise them (can't be copyrighted only digitized rights for digital products). So, unless the ai has produced the written word, verbalised the written word and published the written word copyright ambiguity does not apply.

Depraved wrote: »

you can still copyright the assets of your game (art and sound), the name, the logo...

There is potential that the end result once this all goes through the courts would be that any work with any portion of AI generated asset in the final product can't have a copyright at all.

Yes, copyright is automatically granted once something is released (not once it is written), but only if it is eligible for copyright.

Keep in mind, the company that is wanting this to be the case is the same company that made it so some assets that should have gone in to the public domain in the 1960's have only just recently gone in a few years ago. This company will argue this point, and US courts will side with them over what is best for US citizens.

That said, I'm not saying this is how it will end up years down the line, I am saying this is how it may end up - and it is probably not worth that risk - just allow someone to mod it in.

Depraved

you make a song, you copyright the song, then you put the song into the game. you can do it since you made the song.
you draw a character, you copyright the character, and then you put the character into the game. you can do it since you made the character.
or you can copyright the game and all artistic assets are immediately protected.

adding non copyrighted elements in a video game doesnt invalidate the copyright of other elements because games already contain non copyrighted elements...

also, ai has been in use in video games for decades. even the code to make the ai is copyrighted...what it isnt copyrighted is an artistic work made by a computer, not a human. if an ai makes a painting, or writes some text, etc, but the program that wrote it might be copyrighted.

McShave

Noaani wrote: »

There is potential that the end result once this all goes through the courts would be that any work with any portion of AI generated asset in the final product can't have a copyright at all.

Yes, copyright is automatically granted once something is released (not once it is written), but only if it is eligible for copyright.

Keep in mind, the company that is wanting this to be the case is the same company that made it so some assets that should have gone in to the public domain in the 1960's have only just recently gone in a few years ago. This company will argue this point, and US courts will side with them over what is best for US citizens.

That said, I'm not saying this is how it will end up years down the line, I am saying this is how it may end up - and it is probably not worth that risk - just allow someone to mod it in.

It doesn't make sense to me how having a little bit of AI could remove all copyright laws for a product, even the parts that weren't AI generated. However, there are a lot of US laws that don't make sense to me, so it could be plausible. I guess the technology is just too new and the laws haven't caught up yet. I'm sure by the time this game releases, the legal battles have been played out and we'll see how the land lies.

Great point tho, something to watch out for.

TheDarkSorcerer

It's needed. I can't stand games where NPC's don't talk at all. It's just chat boxes...

Depraved

TheDarkSorcerer wrote: »

It's needed. I can't stand games where NPC's don't talk at all. It's just chat boxes...

so u hated games for the last 40 years?

McShave

Depraved wrote: »

TheDarkSorcerer wrote: »

It's needed. I can't stand games where NPC's don't talk at all. It's just chat boxes...

so u hated games for the last 40 years?

There is such a thing as new gamers. Also, replaying old games after some of the newer games, you start to see some of the flaws (or things that have been improved since).

Depraved

McShave wrote: »

Depraved wrote: »

TheDarkSorcerer wrote: »

It's needed. I can't stand games where NPC's don't talk at all. It's just chat boxes...

so u hated games for the last 40 years?

There is such a thing as new gamers. Also, replaying old games after some of the newer games, you start to see some of the flaws (or things that have been improved since).

ithink he is around my age, not a new gamer xD

but yeah i agree..id say technical and creative limitations more than flaws though.

Noaani

McShave wrote: »

It doesn't make sense to me how having a little bit of AI could remove all copyright laws for a product.

It doesn't make sense to ke either.

That said, it also doesn't make sense that cootright law in the US could go from a flat 25 years to x years after the creators death, argued by a company in order for them to continue tomprofit off of a cartoon mouse for years longer than they should have.

In America, laws (especially around copyright) aren't made that make sense, or that look out for people's interest - they are made to maximize corporate profits.

Companies that have sway over legislators and make money via selling cartoons simply don't want everyone to be given the ability to create a cartoon that can meet their in terms.of quality. Since AI will allow this very soon (non-public AI already does allow this), the only avenue these companies have is to argue that such works can't have a copyright.

Now, again, I have no idea how this will all resolve itself. It may well be that it is only AI generated parts of a work that can't have a copyright - like, this absolutely is a possibility. However, the companies in question are known for using a sledgehammer to deal with issues that could have been resolved using a toffee hammer, so it is entirely within the realm of possibility that it could end up as any work containing any part generated by AI being inelegable for copyright as a whole work.

Child Item