The beginning of a new Era or the final defiliment - The project to give voice all Morrowind's dialogues with AI has begun

[Jarmaro]

The mod author is working tirelessly, we are now at version 0.0.9 and 5.59% of Unqiue NPC dialogues. Support has been added for Hasphat Antabolis, Daric Bielle & many more Nord lines, and also Aryon.
I assume the count for Unqiue NPC dialogues includes factions and optional quests, so we will arrive at fully voiced MQ way before this is over. At the pace it goes on it will be done in less than two months.

 

[Harthwain]

Whats the advantage of listening intead of reading?

Immersion.

 

[Jarmaro]

Here's the tracker for what Dialogue has already been covered. Have you known Torasa Aram, the lady that manages the Museum of Mournhold, has the most lines in the game? Funny.
https://kezyma.github.io/html/vov_progress.html

 

[dreughjiggers]

I used up 20k of my 100k quota for the month tweaking Divayth Fyr's lines to sound passable, and I've still got 2/3s of his dialogue to go. So all these niggers just recording dialogue without giving any ear to their handiwork are full of shit. There's no way it sounds good, in game.

If you actually have a working Divayth Fyr's package, consider sending it to this mod author. He's already included Dagoth Ur and Almalexia from other authors, I'm sure he'd love to see Divayth.

Might as well. I have a lot of his dialogue ready, and Tony Jay as Caius Cosades: https://voca.ro/12OLEpbfQIzg

Whats the advantage of listening intead of reading?

Grow three inches? idk

*acknowledge this users agenda*

The fuck does that mean? Are you ESL?

 

[Bohrain]

Morrowind isn't really the kind of game that benefits this as most of the lines are the shared hyperlink descriptions so few characters have a unique way of talking to begin with. And Bethesda games tend to have few voice actors voicing way too many characters with awful voice direction so next to no one sounds distinct.
The benefits of this is making modded content not stand out in voice acted games and changing or implementing new lines for actual products. Though the latter depends on how the contracts and prices of the selling voice samples for individual production will work out. But I can see that from a practical standpoint this makes the voice asset production a lot simpler. Since one engineer could do this without a strict time constraint, instead needing to rent a studio and hire and schedule an actor and a whole studio crew. Of course this is assuming you forgo motion capping and modeling the likeness of actors, but for single or double A productions that's rarely a thing to begin with.
I don't know how much this will affect the quality of voice acting or how much indie devs get access to it. Quality of voice acting tends to be more about finding competent C-listers to hire and direct them and if people end up using generic AI generated Matt Mercer voice for everything, it'll be as grating as it's now. As for the price, who knows where the contract prices for licencing someone's voice legally will end up in.

 

[Jarmaro]

It's funny how we've arrived in the dreaded point in time when the technology moves so fast we cannot even adapt. By the time any legislation for AI passes, it will long past the moment when it could be applied in a meaningful way. When people talk about AI voices, they only consider the current state of affairs, as if the technology wasn't barely usable just a few months ago. Today it can already be competetive in some areas with humans, and it wasn't even developed in a serious, dedicated way. Commercial applications will speed-up the developement tremendously, what state is it going to be in a year? The same as a human actor, but with varied tones and emotional range avaialable at one sample and click of a button?

It develops faster than a production cycle of an average video game. It's not even worth it using it now, as soon it will be better than it was previously. In just few months AI startes dominating cheap art production and voice-acting. I've seen technology that's supposed to create objects and meshes just from photos. How soon will it be able to do it accurately and consistently?

Funny thing about video games I've noticed, they don't have to even be good to be succesful - they just need not to suck. A game can be perfectly average in almost every regard except for one where it's above average, then it gets called a gem or a classic. A perfectly average (by typical standards of rating) video game can be successful. How far are we from a video game blueprint made en-masse by an AI that is trained to consider all other AIs and plan a game around them? All you would have to do is take a blueprint and start making a game, with all theoretical background already done, more precisely and efficiently than any designer.

I once heard that a book told a story about some kind of post-singularity setting where people had problem with media and art made by an ai - the problem was that AI made so much crappy content that you couldn't find the actually good one and were forced to endure mediocrity, all the time. I guess it wouldn't be that bad with games as we need to endure extreme mediocirty all the time, so if anything there will be an improvement.

 

[Harthwain]

It's funny how we've arrived in the dreaded point in time when the technology moves so fast we cannot even adapt. By the time any legislation for AI passes, it will long past the moment when it could be applied in a meaningful way.

We were way past this point a long time ago.

 

[Kalon]

Whats the advantage of listening intead of reading?

Immersion.

This is a buzzword that has stopped meaning anything and needs to die.

 

[Harthwain]

This is a buzzword that has stopped meaning anything and needs to die.

Uh-huh.

There is a vast difference between reading and listening to a good quality voice cast. I should know, because I have some games as well as audiobooks I listen to frequently. I listen to them mainly when working, working out or commuting, but if a video game gets great voice acting, I am in favor of it.

The real problem is: in this day and age games seldom get good enough voice acting to justify it, so these resources are often better spent on other things. But that's a different position to take from writing off voice acting entirely from get-go.

 

[KörangarTheMighty]

Voice conveys emotional nuance better than text, provided the actors are competent. Silent film went the way of the dodo for a reason. So it goes for games.

 

[Nifft Batuff]

This is a buzzword that has stopped meaning anything and needs to die.

Uh-huh.

There is a vast difference between reading and listening to a good quality voice cast. I should know, because I have some games as well as audiobooks I listen to frequently. I listen to them mainly when working, working out or commuting, but if a video game gets great voice acting, I am in favor of it.

The real problem is: in this day and age games seldom get good enough voice acting to justify it, so these resources are often better spent on other things. But that's a different position to take from writing off voice acting entirely from get-go.

I prefer reading a book than listening it. Book format is not suited for VO. The same applies for many videogame formats.

 

[deuxhero]

Books depend on the writer. Some authors (Larry Correia) are way better for narrated books than others (Tolkien and his nonstop overuse of "said" when it's just two people talking. Why would I need to see/hear "Gandalf said"? I already fucking knew Gandalf was the one who said that because him and Frodo are the only damn people in the scene and Frodo addressed him by name in the previous line. Also Frodo and Gandalf do absolutely nothing else while they're talking.).

 

[Harthwain]

I prefer reading a book than listening it. Book format is not suited for VO. The same applies for many videogame formats.

Bullshit. This may be true for some books, but not for others. I found Conan to be really good in audio format.

 

[Nifft Batuff]

I prefer reading a book than listening it. Book format is not suited for VO. The same applies for many videogame formats.

Bullshit. This may be true for some books, but not for others. I found Conan to be really good in audio format.

Bullshit. Reading is much better. It's more personal and I can immerse myself in the story better with my own pacing. Also it is more fast. When I read a paragraph and understood the whys and hows, the VO is still reading at the beginning of the same paragraph. It destroy the pacing and push me to do something else instead that waiting a stupid VO.

 

[Harthwain]

Bullshit. Reading is much better. It's more personal and I can immerse myself in the story better with my own pacing. Also it is more fast. When I read a paragraph and understood the whys and hows, the VO is still reading at the beginning of the same paragraph. It destroy the pacing and push me to do something else instead that waiting a stupid VO.

Eh, you don't even understand what I am saying. Nevermind then.

 

[Poseidon00]

 

[Jarmaro]

Now that's a good voice for a protagonist. If I read correctly, it's a female Inqusitor voice from SW:TOR.
https://www.nexusmods.com/skyrimspecialedition/mods/84814?tab=description

 

[Perkel]

I can see many things this makes possible:

1. "We can't provide many choices because VA is expensive and you can't make non-voiced games anymore at AAA level." excuse is gone.
2. Indies will have the same ability to make bigger RPGs as bigger studios. AI making assets is just around corner as well.

VoiceAI intonation also is around corner and not as a tool but AI will read other part of text use it as cue and then read "" part of text.

Something like:

She looked at him and screamed with shock: "Marcus ! Come here quick !"

AI will be able to only read "Marcus ! Come here quick !" with proper intonation according to "She looked at him and screamed with shock"

source:

I play with that voiceAI for past 2 weeks and results are pretty great to scary good.

 

[Lagole Gon]

Lmao. This one is good.

 

[Jarmaro]

Hmmm, unsure about this one. On one hand the dialogues appears to be mostly good, on the other the voice...too up-beat? Smug? Too condescending? It does match the casual arrogance of Vivec, but I have always imagined Vivec as calm and professional in his conversation with Nerevarine. The voice is Aradesh for Fallout.
This voice wasn't done by Kezuma btw, Kezuma intends to do Vivec voice based on Vivec's in-game dialoges - those unused ones, as far as I understand.




Diavyth Fyr on the other hand...I think it fits. The greeting would need to be adjusted to sound like Fyr, but it works.

 

[Perkel]

Hmmm, unsure about this one. On one hand the dialogues appears to be mostly good, on the other the voice...too up-beat? Smug? Too condescending? It does match the casual arrogance of Vivec, but I have always imagined Vivec as calm and professional in his conversation with Nerevarine. The voice is Aradesh for Fallout.
This voice wasn't done by Kezuma btw, Kezuma intends to do Vivec voice based on Vivec's in-game dialoges - those unused ones, as far as I understand.
Diavyth Fyr on the other hand...I think it fits. The greeting would need to be adjusted to sound like Fyr, but it works.

Method they are using is to get 4-5 10 second samples and then AI spits voice in vein of those 4-5 clips and CAN emulate raising voice etc.
The problem is that if you feed it someone talking in speech manner then voice will sound like it was a speech not natural talk.

Whole thing is pretty developing fast so maybe in next month or two you will be able to feed it more lines and like i mentioned before give voice a context from which it will change tone.

 

[Jarmaro]

This is just out of this world.

 

[PapaPetro]

If a fucken robot can voice lines better than you then you're doing a terrible job at being alive.
No tears for those that 'phoned in' their soul.

 

[Perkel]

This is just out of this world.

 

[Lagole Gon]