Jabberwocky: Sims vs. KOTOR / Jade Empire

I think what killed KOTOR (speech wise, it is still a great game in all other aspects and one of the most engaging RPGs I've ever played - the sequel much more so than the original) was that they pretty much recycled two or three loops for everything the non-human races said. That made it annoying to listen to, so you'd be clicking to continue as soon as you finished reading, which kind of killed the immersion factor. It did sound like Star Wars, though.

Personally I'd rather not see any made up gibberish in a game. I'd much rather hear real languages, like the Nepalese in recent Uncharted 2. Even though parts were obvious from context, I didn't actually understand a word that was said (and there are no subtitles for those parts) but I enjoyed every moment of it. But for a science fiction setting this might not make any sense. If they really have to create a language for it, they should make sure to create enough so that everything they want to convey can be expressed in it. Then I'm sure I wouldn't feel so annoyed listening to it. Those loops are just a cop-out.

I like this:

Suppose that as a character speaks, their speech is printed on screen either word-at-a-time or letter-at-a-time.

For the sound of their voice, the character has a small (5-10) library of vocalization samples for each of their moods (each mood has a different library). For human-like voices a good sound might be 'blah' for conversational, or 'bleugh' for drunk or 'gab' for gossiping or 'hrumph' for irate/angry or 'YARRR' for pirate.

A sound from the relevant library is played either every few letters (or every letter if they are short) or every word, or something. Each sound is selected randomly and is distinct from the previously played sound.

One of my favorite features of this approach is that for non human characters you can use very unusual sounds to convey the sounds of their voices. Computers can make bleepy sounds, aliens/spirits can make really weird sounds. Etc.

I haven't played KOTOR or Jade Empire, I'm even further behind. But what appealed to me about the Sim babble language was the way that it enhanced the thought bubble icons that would be displayed when the communication was occuring. You were able to take those two elements and piece together the jist of the conversation. However, if for some reason, I needed to know exactly what the conversation was about, that babble would be quite pointless and counter productive.

If I were to take a run at using made up speach, I might apply it to unimportant NPC citizens wandering around to give them a little more flavour. I'd want to include that icon in a bubble or provide something else to indicate a generic topic. The result should be like if you were sitting at a bus stop listening to a couple people talking. You don't really know what their conversation is about but you can get the basic idea. It's not relevant to you in any way and you don't really care what they're saying but it makes sitting at the bus stop more interesting than if everyone was silent. That the characters in the game would be speaking in gibberish would (hopefully) make it clear that their conversation is not relevant to whatever quest you're on.

Generally, I like speech that's either english (or whatever the player's native language is) or deep. In a sci fi setting, for example, there are two options for me:

1) Make up a reason that the aliens speak english, like a translation device

2) Make the languages real, like Tolkien did.

The reason gibberish languages are annoying is because they are obviously fake -- they are just sounds without any information in them. If you want an alien language, create a language with real grammar and let the aliens speak it.

Of course that's a huge undertaking, so I think most people who aren't obsessive linguists should just go for option 1.

One aspect that makes Simlish work is that it was done by skilled Improve comics. They were there together in the recording studio and playing off one and other, basically replying to each other in a made up language.

Simlish is not a language, it is an expression of emotion. And they expressed it very well.

Quote:

Original post by lightbringer I'd much rather hear real languages, like the Nepalese in recent Uncharted 2. Even though parts were obvious from context, I didn't actually understand a word that was said (and there are no subtitles for those parts) but I enjoyed every moment of it. But for a science fiction setting this might not make any sense.

If I remember correctly, many of the random phrases and background chatter in FireFly/Serenity was in Mandarin. That seemed to work very well - at least for all us Westerners who don't understand Mandarin...

Oh, I guess I didn't actually weigh in on the question; I like Jibberish, you might loose a little bit of the inflection in well acted speech, but I think you can still successfully communicate the character of the speaker, and his/her mood.

Another nice feature of using jibberish is you don't have to localize the speech.

Animal crossing is another game with irritating fake language. I think what makes sim language work is that it does actually convey information, because it conveys emotional tones relevant to what's going on in the game: compliments sound complimentary, swearing sounds frustrated, lack of approval sounds dubious, laughter and crying are laughter and crying.

Spoken text in games I find is the same as another game feature. So if you can’t do it well than you’re better not including.

Sure the game won’t be as vibrant without spoken dialog but people are more forgiving of not having it than badly done voice acting. Like wise half done voice acting is just as annoying. I was playing an RPG not long ago that only had voice acting for key characters and their plot specific dialog. Which was jarring ask them about the killer and you got a couple of lines of text and dialog ask about their life story and you got a couple pages of text but no dialog.

Is there a way to make voice procedural just as you make graphics procedural? That would be a cool way.

(if procedural (on-the-fly rendering/creating) is the wrong term, please correct)

Quote:

Original post by Alpha_ProgDes Is there a way to make voice procedural just as you make graphics procedural? That would be a cool way. (if procedural (on-the-fly rendering/creating) is the wrong term, please correct)

There are several computer programs which will read written text aloud, but they can't do emotional inflections anything like a real voice actor. Even a mediocre voice actor is better than a synthesized voice.

Quote:

Original post by sunandshadow Quote: Original post by Alpha_ProgDes Is there a way to make voice procedural just as you make graphics procedural? That would be a cool way. (if procedural (on-the-fly rendering/creating) is the wrong term, please correct) There are several computer programs which will read written text aloud, but they can't do emotional inflections anything like a real voice actor. Even a mediocre voice actor is better than a synthesized voice.

True, but we're talking about a made-up language (ex: glub, blah, foo, bleh, blep). So it should be possible to get inflection randomly thrown in there, somewhere. You really wouldn't get a voice actor for that. It's not like they are speaking Klingon.

The gibberish languages that I find most tolerable are the ones that are relatable in terms of inflection. I want each utterance sound like an English speaker who's just incoherent, with each sound coming across as inquisitive, delighted or angry. If you can get that much, and keep the utterances short (The Sims never hold long sililoquies, it's just a blurt to go along with their thought balloon, whereas Midna in Twilight Princess would go on forever in her little imp voice), then I think it'll be okay.

Quote:

Original post by sunandshadow Quote: Original post by Alpha_ProgDes Is there a way to make voice procedural just as you make graphics procedural? That would be a cool way. (if procedural (on-the-fly rendering/creating) is the wrong term, please correct) There are several computer programs which will read written text aloud, but they can't do emotional inflections anything like a real voice actor. Even a mediocre voice actor is better than a synthesized voice.

This of course is assuming you have them reading plain text English.

If your speech synthesizer is instead reading "Directed Script Language", a special markup over the English Language, then I don't see why you couldn't Write all the extra stuff a good voice actor would add.

Quote:

Original post by Talroth Quote: Original post by sunandshadow Quote: Original post by Alpha_ProgDes Is there a way to make voice procedural just as you make graphics procedural? That would be a cool way. (if procedural (on-the-fly rendering/creating) is the wrong term, please correct) There are several computer programs which will read written text aloud, but they can't do emotional inflections anything like a real voice actor. Even a mediocre voice actor is better than a synthesized voice. This of course is assuming you have them reading plain text English. If your speech synthesizer is instead reading "Directed Script Language", a special markup over the English Language, then I don't see why you couldn't Write all the extra stuff a good voice actor would add.

Direct SL... you may have given Microsoft their next API!

I think the most important points of non-language speech is inflection and randomness. You don't want the same flat, repetitive track over different sentences such as: "I took a crap." and "He went to fight the Duke... alone!"