Rumors are flying around about the next generation of consoles, and that they're rumored to be extremely powerful. PC's are more powerful than ever of course, but the advancement in consoles means that developers can finally raise the bar again, because they won't be tied down to the lowest common denominator anymore.

So anyway, it's time to talk about a feature that we will begin to see in video games in the near future.

The Future of Video Game Dialogue

Right now games like Skyrim and Mass Effect have dozens of hours of spoken dialogue which must be stored on the game disk. However, we will soon see procedurally generated voices and dialogue in video games, mostly in RPGs of course. Don't worry though, genuine voice actors will likely still voice important characters.

In the next generation of games you'll be able to actually talk to NPCs. Microphone (and Kinect) support will be fully utilized, so the player can speak, and the NPCs AI will be sophisticated enough respond to your voice. They will actually be able to put together a full conversation on the fly. Additionally, voices will be unique to each character, and conversation will draw from phrase dictionaries, and use logical pathways to construct meaningful lines. The game will be able to generate very life-like speech patterns and accents, these will be coupled with a variety of filters (i.e. pitch, throatiness, etc.) to give distinct tones to each character.

Early titles that incorporate these features may sound a little funny at times, and say some weird things every so often. Eventually though (in less than 10 years), this artificial dialogue will be nearly indistinguishable from real voice actors. And even if it does sound a little off sometimes, it's still a lot better than hearing a certain NPC repeat the same line for the hundredth time (I'm looking at YOU Belethor!).

Don't be surprised to see this handled by sophisticated middleware. I'm betting that dialogue toolsets will be like the next Havok or SpeedTree. If Bethesda is smart they'll develop this middleware themselves and license it to other publishers.

Post your thoughts below.

I would just be happy if, for once, MY character would talk, instead of just picking options from a menu and delivering my lines silently. All this future stuff can stay in the future, I'm happy with the current set up, the voice overs and what not. If they could invest time into actually making the player character talk (like they do in some games - DA 2 for example) then that's enough for me, for now.

That would be amazing, if they can record so much voice acting to give us voice choice for every race and six. Huge resources needed.

Silent protagonist vs talking protagonist have been a big issue in Bioware forum.

Some players prefer silent protagonist and so they can role-play the character, i my self prefer silent protagonist.

Some players want talking protagonist with their own reason.

in my personal opinion, talking protagonist didn't make me feel being the character. That character is someone else, not me being her/him. It is like watching a movie rather than being in the movie. Secondly, the resource for the protagonist voice (sound file) better contribute it in other things such as better graphic, better visual, better game mechanic, better sound effect, no bug and no glitch or anything to improve a game.

"Talking" to NPCs in the game is emotional, voice actor for the protagonist didn't give that. The player talk in their imagination when choosing the line. Voice actors don't represent the player emotion.

You're right. Resources is a big problem. Even compressed 50+ hours of spoken dialogue takes up a huge chunk of disk space. Imagine another 50+ hours of player dialogue in multiple voices.

I would very much welcome any future development or technical breakthrough that removes the limits that full voice acting places on dialogue: disk space, inability for small updates and changes late in development, voice actor costs, and so on.
Full voice acting, while probably unavoidable for a major title like TES games, is IMHO the main culprit of the decreasing quality and, especially, amount of dialogue in games. Morrowind prides itself with "dialogue equivalent to several novels" (or words to that effect), and that is a thing of the past.
Fewer dialogue means more shallow writing, less emotions, less characterisation, less involved quests - the essence of a role playing game.
So if we could have voice acting without the technical and financial constraints currently imposed, would be really great.

This is exactly why the next generation of games will be adopting this method.

I hope they keep the voice acting for the basic dialogue lines (about as long as Skyrim's) and add on top of that some extra text-only dialogue (like Morrowind's) for the players who want to go deeper and find out more about quests, each npc's background, family, trade, culture, religion, etc, quest and non-quest directions, rumors.... I'm sure those people who want heavy branched dialogue won't complain too much about the lack of voice acting in those additional discussion topics, after all the essential dialogue is still fully voiced.

That would be fantastic, but I doubt it will happen anytime soon and particularly from Bethesda. They aim high with all these "radiant" technologies but their games are losing the cohesion touch, they can't even add a dialogue condition check like [if you already met the player and already told him phrase no.114 about working at Belethor's, then goto phrases (115 to 118) and pick randomly (or based on disposition) a generic hello, nice to see you or [censored] you or just be quiet for a change]. This is not rocket science, unlike the player-npc actual talk simulation.

That would kill quite a bit of role-playing. I choose how my character sounds, not a VA.

There would have to some major breakthroughs in chatbot technology. The most advanced chatbot been worked on for years and its still far from being remotely realistic. Imagine thousands of those chatbots with completely different personalities. It might actually be more work than recording the lines from real people.

It's all about the money. Chatbot software doesn't have a very big market right now. When it becomes a feature in video games you'll see a lot more money being put into it's development.

And like I said, it will likely be handled through middleware...
So the the cost to develop it could be offset by lucrative licensing deals. Lets say that EA was the first to the market with it, they could opt NOT to license the toolset to their biggest competitor: Activision. Then EA could go around touting that it has this technology, while Activision doesn't. Publishers are shady like that.

It wouldn't really be too difficult for Bethesda to implement a "Mute Player", or a "Player Dialogue Volume" option in the pause menu.

We probably see this kind of dialogue tech with the real big boys of game technology like Cryteck, DICE, etc. Beth is a successful solid game developer but they aren't ambitious enough. Cryteck has a whole department dedicated to the research of graphical tech and whatnot. They invented and use for the first time in a game SSAO. Procedural dialogue will be a interesting thing to create.

Id like to just throw in there that dragon age origins had a nice selection of different dialogue accents for the charector to pick from. Where as I would like to chose and hear my toons voice when in speech dialogues, I would not like him sprouting all this different type of rubish while fighting and so forth like in the that game.

Yes, that should be toggleable in the audio menu because it's getting annoying fast.

Well, I think I'm much more in favor of the silent hero. The silent hero gives you more room to react yourself, and it adds to the possibilities of choice.

It adds to choice because a studio could only afford to have a few choices of interactions and even then not on every quest. So you'd be forced into perhaps 2 responses and that would be it.

If you had a quest to investigate something corrupt, you might have only two options -- absolute outrage and absolute indifference. And the answers would probably be one line (remember we must account for all 9 races and both genders, so having two answers at this point means having 36 recorded responses). So keep them silent so I can make up one of thousands of responses.

There's supposed to be a very sophisticated response program out there (developed with or for government agencies) that will mimic a person, and if the conversation doesn't go on too long, or you are not too cryptic, you will think you are speaking to a real person. If this really is already out there, then, like most technologies, it will be co-opted for the purposes of entertainment.

No doubt. I'm sure that games will have constraints and limits to what the NPCs can understand.

If you walk into a Weapons Shop and ask the shopkeeper if he sells the new iPhone, he'd say "What is this you speak of nave?" or "Sorry, I don't know what that is." The word "iPhone" would be omitted from the games vocabulary of course, so the AI wouldn't have anything to go off of.

The next best thing is to have NPCs respond to keywords you say. That way, chat bots don't need to be involved while still maintaining the illusion of talking to someone.

Chatbots have to work off of keywords anyway, and their responses are limited by that. They can't "learn" new dialog that hasn't been given to them in advance; all they can do is decide which of their responses is most appropriate. I've seen some amusingly bad attempts at it so far, and you're talking about a major step beyond that.

Considering how badly most NPCs tend to judge your character in TES, and base their dialog on that, I can't see anything good coming from this in the near future unless Bethesda does a major about-face in its game designs. DF and MW at least took some notice of your achievements and factional standings in a few cases, and offered SOME appropriate choices of response here and there, which was possible due to the majority of it being text-based, but that all but vanished with full voicing in OB. From the comments here, I'd tend to believe that Skyrim is even worse at it.

Like I said, this hasn't taken off yet. We'll see this technology advance very quickly. If you haven't done so yet, go demo Siri.

What I'm talking about is a system that works similar to Siri. But it would be different, because instead of typing emails for you, or looking up the weather forecast, the NPC "Siri" would go to a word and phrase database. Algorithms would then determine the best response to reply with.

Each NPC would have to be customized to determine the extent of their knowledge, and the potential things they can say of course. I.E. You couldn't go up to the Jarl's cook, and ask her where to find ore deposits. Likewise, a weapon smith wouldn't have access to phrases involving illusion magic.

Chatbots is one thing, realistic human speech is another. Currently, speech synthesis can only express one emotion, which is simply not good enough. One voice could be about 500 MB of samples. With 100 actors, thats 50GB just for voices. We are talking way far off into the future, here.

No not at all. There won't even need be actors or recorded dialogue samples for the basic NPCs. It will go from text to synthesized speech. I said:

"The game will be able to generate very life-like speech patterns and accents, these will be coupled with a variety of filters (i.e. pitch, throatiness, etc.) to give distinct tones to each character."

That's why I started the OP talking about computing power. This synthesized speech will be relatively processor hungry.