Return to “Suggestions”

Post

Re: Voice Acting - Crowdsourcing?

#196
BFett wrote:I would definitely record some voice snippets for LT if I was given things to say..
: generic bordom x 20
: going to die x 20
: finally found you x 10
: going to collect that bounty x 10
: swearing, where did that come from x 20
: Oh god im getting outta here x 10
: generic cussing x 10

... :V
yeah we could do this.

: alpha through zulu (variations of each)
: zero through niner (variations of each so you dont have the same letter three times sounding horribly repetitive :V )
: Transporting/Hauling variations x 5 eg "Im Hauling", "Im transporting", "im carrying"
: heading to variations x 5 eg "headed to", "going to"
: bounty hunting starter x5 "Im hunting"

Easily possible
°˖◝(ಠ‸ಠ)◜˖°
WebGL Spaceships and Trails
<Cuisinart8> apparently without the demon driving him around Silver has the intelligence of a botched lobotomy patient ~ Mar 04 2020
console.log(`What's all ${this} ${Date.now()}`);
Post

Re: Voice Acting - Crowdsourcing?

#198
Talvieno wrote:Definitely! Random place names, though, is a no-go. Even trying to guess syllables would be a total pain.

I do hope Josh isn't still using his old name generator code though... :?
IF we have a name generator based off of syllables then we can record some variations of each syllable.
Then it can piece them together like FL did for entire sentences. only with smaller gaps.
°˖◝(ಠ‸ಠ)◜˖°
WebGL Spaceships and Trails
<Cuisinart8> apparently without the demon driving him around Silver has the intelligence of a botched lobotomy patient ~ Mar 04 2020
console.log(`What's all ${this} ${Date.now()}`);
Post

Re: Voice Acting - Crowdsourcing?

#200
Talvieno wrote:But that's yuck! :V That's hardly anywhere near infinite names. I want an infinite name generator. I'll supply the namegen code. :D It will be a mod.
With pronounceable syllables and reasonable length words you can only have limited names.
°˖◝(ಠ‸ಠ)◜˖°
WebGL Spaceships and Trails
<Cuisinart8> apparently without the demon driving him around Silver has the intelligence of a botched lobotomy patient ~ Mar 04 2020
console.log(`What's all ${this} ${Date.now()}`);
Post

Re: Voice Acting - Crowdsourcing?

#202
Talvieno wrote:But enough that you'll never see the same two twice! Taiya's word generator is capable of generating over a million completely unique, fully pronounceable words in one go. For IRC reasons I limit it to 20 or so.
If they are pronounceable, then there are syllables, so we can record each unique syllable, and hey presto we have pilots speaking the random names :D
°˖◝(ಠ‸ಠ)◜˖°
WebGL Spaceships and Trails
<Cuisinart8> apparently without the demon driving him around Silver has the intelligence of a botched lobotomy patient ~ Mar 04 2020
console.log(`What's all ${this} ${Date.now()}`);
Post

Re: Voice Acting - Crowdsourcing?

#204
Silverware wrote: We could work with categories,

Transporting Raw materials,
Transporting electronic goods
Transporting, none-of-ya-business!

Place names will be a pain, but if we work with the name generator and do syllables, we could generate a reasonably large number of potential names of places/pilots/ships/things

The pilot-to-pilot chatter will of course be the easiest, and yes, hundreds of thousands of voice clips to get this right is important. :D
But then, it would add so much to the game to have people's voices and such.
Being able to pronounce those randomly generated placenames would require that we have all our possible syllables pre-recorded, which imposes some logistic problems, so you're asking for trouble with syllable generators. For some languages it's not too complicated, so we'll start with Hawaiian:
  • Vowels:
  • short: [a e i o u]
  • long: [a: e: i: o: u:]
  • Consonants
  • [m n p k ʔ h w l]
Syllable structure: (C)V.

Hawaiian's syllable structure and phonotactics are so trivial that this is a basic combinatorics problem. We have ten vowels (five long and five short) and eight consonants.

(C)V means we allow syllables of consonant + vowel as well as syllables consisting only of a vowel.

We have consonants * vowels + vowels to cover CV + V, so we end up with 90 possible syllables. In reality, this example turns out to be reasonably accurate, where /wu:/ is the only illegal CV syllable in Hawaiian. But it gets complicated. In addition to having unit vowels, Hawaiian also has diphthongs, which are vowel sounds that contour from one vowel quality to another (cf. the vowel sounds in English "ouch", "night", and "Friday" -- both of them).

Hawaiian has about 15 of these diphthongs, and a diphthong can take the place of a vowel in a syllable (they are both valid nuclei).

So now we can consider D as our number of diphthongs and set up our math problem this way:

(CV + V) + (CD + D) ==> (8 * 10 + 10) + (8 * 15 + 15) = 90 + 135 = 225 possible syllables.

This might not be entirely accurate as far as actual Hawaiian words play out. In reality, a lot of potential syllables are never used, especially in languages with very permissive rules. English allows many forms of CCVCC, but indeed, "swict" is not a lexical item outside the paraphonological realm of acronyms and Twitter hashtags. So while 225 possible syllables isn't necessarily an accurate depiction of the sonic depth of Hawaiian's lexicon, it's definitely good maximal case.

So if Hawaiian were to feature in Limit Theory, we would need up to 225 sound clips per voice actor. In English, we have about 10-14 vowels and 24-25 consonants depending on dialect. Just covering possible (C)V syllables, we're dealing with a maximal case of 364 syllables, and that's not even getting into the rest of English's (CCC)V(CCCC) syllable structure. Thankfully, maximal or near-maximal words like "strengths" are fairly rare and not every single legal combination of CCCVCCCC would need to be pronounced -- just the ones that are actually used in the English lexicon.

Though with a lexicon of well over 100,000 words, there would need to be some trimming done. But if, say, we limited LT's lexicon to the ten hundred most common words or the two thousand most relevant words to spacefaring, then we wouldn't even need the syllable complication and we could record whole words, though at that point, we might be back to how "blocky" Freelancer's NPC dialogue sounds. I suspect we'd have similar issues with syllables though, so perhaps I got on my linguistic soapbox for naught.

If you want to go further down this rabbit hole, reveal the spoiler below.
Spoiler:      SHOW
Limit Theory, though, is procedurally generated. It doesn't have to stop at one language. It doesn't have to stop at a dozen languages, or a gross of languages, and those languages could have any set of phonemes and any list of phonotactic rules that restrict and permit various combinations. But let's stop and do an intermediate kind of check -- barring other structures, what are all the possible (C)V syllables that could conceivably be used in human languages?

To make this a little less complicated, I'm going to discount diphthongs and secondary articulations on consonants -- we're just dealing with plain, unadorned consonants present in the IPA table, and plain unadorned vowels in the IPA table. So here we go:

We have about 130 consonants between pulmonics, ejectives*, implosives*, and clicks. We have 39 vowels in the standard vowel table**. Per our CV + V formula, we get a whopping 5109 possible syllables. If we considered additional details about articulation, this value would increase at least one order of magnitude. If we added more types of syllables (not everyone speaks Space Hawaiian!), then we start entering "is this pronounceable" territory.

Georgian, for example, is a language that permits a maximal syllable of CCCCCCCCVC structure. That's eight consonants in the onset of a syllable.The most we permit in English is three, for comparison. Nuxalk defies the entire notion of a syllable itself, permitting entire words consisting of nothing but consonants.

It doesn't take too long for us to want to break things into smaller units than syllables, or to limit all of our potential procedural languages to a specified subset of different human sounds. Big enough to produce variety, small enough to keep it pronounceable. And making syllable structures and phonotactic rules very restrictive, to keep the overall syllable count down.

* Ejectives and implosives are marked with diacritics, but are distinctly different from the pulmonic consonants at their place of articulation. Either way, this is an order-of-magnitude sort of experiment we're dealing with here, so exact numbers aren't super important.

** The IPA vowel chart uses some diacritics to fill in spots in the vowel table. I am counting these, but I am not counting vowel qualities based on length, tone, or other things. Merely height, backness, and rounding.
Shameless Self-Promotion 0/ magenta 0/ Forum Rules & Game FAQ
Post

Re: Voice Acting - Crowdsourcing?

#205
I know Josh enjoyed the sense of life the general background of chatter provided for Freelancer and I never tired of some cocky pilot telling me what he/she was going to do to me in the fights. I understand it is not easy to get it right but Freelancer managed to do a fair job of it all those years ago. If pure procedurally generated games are going to be taken seriously by gamers it's something that has to be tackled sooner or later.

It was Josh who said he would have a word with his sound people to see what was possible for LT and there were genuine offers of help from members of the Community. :D
As far as I know the coding warrior never definitively said it wasn't going to happen. Text based dialogue is okay..ish but it's a poor substitute for the real thing.

Some clarity on this issue would be appreciated by me. Maybe it's part of the surprise of untold details, one of the "secrets" of LT which Josh is keeping to himself.

I moved past purely text based dialogue a long time ago. I can't imagine a game with universal appeal ever resorting to purely text based interaction. SC would be much poorer in content without its voice acting. :angel:
Post

Re: Voice Acting - Crowdsourcing?

#206
Victor Tombs wrote:I know Josh enjoyed the sense of life the general background of chatter provided for Freelancer and I never tired of some cocky pilot telling me what he/she was going to do to me in the fights. I understand it is not easy to get it right but Freelancer managed to do a fair job of it all those years ago. If pure procedurally generated games are going to be taken seriously by gamers it's something that has to be tackled sooner or later.

It was Josh who said he would have a word with his sound people to see what was possible for LT and there were genuine offers of help from members of the Community. :D
As far as I know the coding warrior never definitively said it wasn't going to happen. Text based dialogue is okay..ish but it's a poor substitute for the real thing.

Some clarity on this issue would be appreciated by me. Maybe it's part of the surprise of untold details, one of the "secrets" of LT which Josh is keeping to himself.

I moved past purely text based dialogue a long time ago. I can't imagine a game with universal appeal ever resorting to purely text based interaction. SC would be much poorer in content without its voice acting. :angel:
This :thumbup:
Post

Re: Voice Acting - Crowdsourcing?

#209
BFett wrote:Well, as with most aspects of LT, if it's not in by default the community can pitch in and make it work as a mod.

(We would need someone to help with adding some noise to the audio files so that we get the effect of a HF radio which is slightly off frequency.)
you have weird digital radios, with perceptible noise and stuff.

your smartphone must be crap :lol:
(or your personal installer of teamspeak/skype/etc)
Post

Re: Voice Acting - Crowdsourcing?

#210
Cornflakes_91 wrote:
BFett wrote:Well, as with most aspects of LT, if it's not in by default the community can pitch in and make it work as a mod.

(We would need someone to help with adding some noise to the audio files so that we get the effect of a HF radio which is slightly off frequency.)
you have weird digital radios, with perceptible noise and stuff.

your smartphone must be crap :lol:
(or your personal installer of teamspeak/skype/etc)
No, I didn't say digital radio. I'm talking old school, High Frequency analog radios which have much more range than your smartphone will ever have. Can your smartphone make a call to the other side of the world without a cell tower network or satellite? No, I didn't think so.
Image

Online Now

Users browsing this forum: No registered users and 20 guests

cron