Check out this new AI speech app (for FMs)

[Azaran]

Because of the obvious risks of misuse

I could envision a dishonest FM author claiming they found a lost cache of original, unused Thief audios from LGS. Not that we currently have any people like that now in the community, but there's always a chance some newcomer could try it

[Kubrick]

I (and I'm not the only one) have been trying to narrow down a consistent acceptable ethical use of generative models

I'd be very interested in discussing your thoughts on it, or any discussions you've found interesting. Personally I am against AI generation in general not only because of obvious possibility of abuse but also because the quality is clearly inferior. It'll usher an age of mediocre media made by producers instead of artists, which is in part what the SAGA strikes are about in Hollywood. The content being pumped by those than own the distribution medias already feels like it was created by algorithms only. Souless and boring, churned out in a digital conveior belt.

But I think I'm okay with it in cases like FMs using Garrett's voice: no one is being more hurt than if you took some of the textures and made similar new ones with the use of an AI model. But those are my initial thoughts on the matter. I'd reconsider in an instant if there was a chance Stephen Russell was back on the job, for example.

I could envision a dishonest FM author claiming they found a lost cache of original, unused Thief audios from LGS. Not that we currently have any people like that now in the community, but there's always a chance some newcomer could try it

I can't even see why someone would make such a claim. I think the misuse is the same that it is in all of AI generation: 1) deceit, 2) put people out of work, 3) lower quality of work, 4) empowering producers and capitalists instead of artists

[gamophyte]

For me I think for a few lines, for this fan audience sticking with FM's, isn't a big deal. I will likely use it for Garrett. I also didn't mind it in Towers of Twilight, so there isn't the jarring break between the conversation then the AI reaction to my presents.

That said, I wouldn't fault anyone who didn't want to use, or be tied to, projects that use AI. Whether it was just for a small audience it still is kind of "bad vibes" when you think about how you're actually hearing a real actor say your lines.

Anyway I appreciate a thread to discuss this as well. I tried to bring up the bad vibes of using original art / textures from other games on the discord and the mod there said shut up kid.

[Kubrick]

There's quite some discussion on the issue already if you search for "AI generated" on the dromed discord, specially so the same topic doesn't gets brought up again and again. Many folks here kinda already made statements on how they feel. The feeling I have is that some old schoolers that really know their way around Dromed, like vfig, are terminately against it while some younger aren't. Some people, me included, see it as just a tool but one with a potential for damage, specially when you're taking someone elses voice. Totally respectable positions, I see valid arguments for both sides.

I'm kinda torn myself really, I feel like it's a great opportunity to breath more life into FMs. Some people make it sound like you just ask some voice actors here and that it, while it's clearly not the case. First off, you actually have to get them, coordinate, ask for reshots (I've done voice recordings and foley sound IRL). Then, we've all heard in FMs how most of the time they sound very different; sometimes the sound texture doesn't match up with the original voices, and even when it does, the guards sound different just for the conversation you produced for the FM then go back to their originals. The small usage in Towers of Twilight, I think, show the best qualities the tool has to offer.

On the other hand, Russell has declined to voice FMs in the past. That was probably simply a carreer desition, maybe he just didn't feeling like it, but his reasons are ultimately irrelevant: He does not want his voice to be used other from what he already has recorded. Hearing that fact made me feel differently about using his in particular.

The problem with AI voices is the potential damage you can do and the ethical considerations of taking someone's control away. In this small modding community, the possibility of damage is incredibly small. Nor Russell nor anyone could possibly be damaged by a few lines in an FM that will be played by, what, 100 people? Regardless, me personally would not feel confortable using his at this point for a FM I made. But it would not deter me to play one that has. It's a spiny subject on development and my opinions may change one way or another.

btw does anyone know if Russell ever made any statements about his voice being used in FMs? I don't remember something about it being talked about directly in the Inside at Last episode, just to name a possible source. Also I don't know where he declined to do FMs. I'd like to read it if it's available.

[vfig]

I oppose the use of AI technology to replace the creative work of individuals, and refuse to be associated with it.

i am restating my position here from the towers of twilight thread, and will provide a little more background (describing voice acting specifically here, but the same applies to any creative medium). my objections stand on these grounds:

plagiarism: use of AI generated voices or voice-changing derived from the work of a voice actor without their consent shows disrespect for that individual and for the creative effort they put into their acting. it doesnt matter if they are a professional actor or not: i reject the suggestion that e.g. plagiarising Stephen Russell's voice would be crossing a line, but plagiarising Mike Chrzanowski's voice is okay. this is fundamentally about you as a creative individual having respect for other creative individuals. (and, of course, much of the best loved voicework in thief was done by non-professional actors, like Dan Thron and Terri Brosius). however, i do not object if you want to use AI generation on your own voicework; or with the voices of people who have explicitly chosen to make their voice samples available for this purpose; or if you want to use the voice of long-dead people (as a rule of thumb, if their voice recordings are in the public domain by virtue of death-plus-X-years laws, then do what you will).

community: nearly every fan mission ever made is from the contribution of many individuals: custom textures, objects, meshes, sounds, artwork, writing, voicework, translations, scripts and script modules from many different people are a crucial part of all but the earliest and simplest fms. we have a small but ridiculously talented community here, and engaging with other members of the community to contribute to fan missions in whatever form builds better missions. it develops skill in the community. and it builds connections among the community that encourage more creative work from all participants. of course there is a little organisational overhead of course in wrangling the work of several people, but hardly as onerous as Kubrick suggests.

creative development: the output of a creative process is only one part of the value in that process; going through the process builds your own creative skills, so that you get better at what you do. and very often, different approaches to the material are only discovered during the process, that make the resulting work far better than what was conceived beforehand. i expect most fm authors are aware of this in regards to their own work; but naturally this applies with all forms of creative work. for instance, i got some adlibs by McTaffer in his voicework for Making a Profit that were a great addition to what i had written in the script (although sadly i could not find a place to fit them in time for the deadline, and had to leave them out). by farming out creative work to automated processes, you are losing this effect, both in terms of the benefits it might provide for your specific mission, and in terms of the individual further developing their own skills.

theft of work: this is not applicable to fan mission creation, since we are all doing this without financial compensation, but i mention it here because it is a separate grounds that underlies my objections in other contexts. using AI generation as a substitute for employing a creative professional is theft of work, and should be opposed by all creative workers.

i think this describes my position adequately. i reject the idea that it is okay because "people will do it anyway"; that others disrespect to their fellow creatives by doing this does not justify me disrespecting them. i reject the idea that it is okay because "i am not taking money away from anyone"; theft of work is only one of the ground for my objections. i reject the idea that "it is too hard to get other people to contribute"; you learned to use dromed, so you dont get to use "too hard" as an excuse for anything else.

i dont understand how any creative person can disrespect the creative work of others like this (and as it seems to me by implication, ultimately disrespecting their own creative work). but realistically i know some of you will be using AI generation anyway. so in terms of those fms: i ask that if you use AI generated works (in any medium, not just voicework) that goes against my objections above, that you do not use any of my objects or scripts in it.

my personal policy regarding such fms: i will still play them, as of course there is a huge amount of original creative work that goes into any fm above and beyond the objectionable content, and i still value those other portions; but i will not post recordings of them to my youtube channel. (i do not expect anyone to care much whether i do or do not post fm recordings; i am not trying to influence anyone into doing or not doing anything by this, but rather i am simply not happy posting such to my channel.)

[New Horizon]

On the other hand, Russell has declined to voice FMs in the past. That was probably simply a carreer desition, maybe he just didn't feeling like it, but his reasons are ultimately irrelevant: He does not want his voice to be used other from what he already has recorded. Hearing that fact made me feel differently about using his in particular.

He's a professional actor and there are union rules surrounding unpaid work. He wouldn't be able to do unpaid FM work without violating the actors union rules and facing a fine.

I personally don't have a problem with FM authors creating new lines for Garrett using AI since they are not being used to generate income. Now when companies begin using it to avoid paying working actors / technicians, that's when I had an issue. Nobody is getting paid for FM's and nobody profits from it. We have Garrett impersonators who do impressions of Stephen's performance, even some of my own voice work has been imitating Benny the drunk guard. As long as it's unpaid, I see no harm.

[Kin]

............................................
I personally don't have a problem with FM authors creating new lines for Garrett using AI since they are not being used to generate income. Now when companies begin using it to avoid paying working actors / technicians, that's when I had an issue. ..............................................

This.

[peter_spy]

@vfig - it's really nice to see someone with such high integrity and understanding of creative processes