So a while a go, a few coders wanted me to try and recreate the AI announcement voice, so they could have more alerts for things. I'm having some difficulty coming recreating the effect. (No idea what they used or how they even did it). So we got to talking for a bit about redoing the AI voices, but never really settled on a conclusion. So i was thinking about running it through you guys first.
So, here's what we'll get for new voices:
Pros:
more announcements
multiple voice types, that could be chosen by the AI player, this includes VOX
feminine
Masculine
Angry
nerdy
All your base are belong to us
Just general higher quality sounds
Cons
We'll lose the current AI alert Sound
more stuff to download
OH GOD IT'S CHANGE, KILL IT
It'll be a lot of work (for me)
/TG/ First and Only Sound guy
The only Dev unanimously loved least hated.
i can convert the Half-life 1 AI voices to .ogg if we really want to go there. But i still like the idea of variety of voice personalities the AI can chose from.
/TG/ First and Only Sound guy
The only Dev unanimously loved least hated.
Cuboos wrote:i can convert the Half-life 1 AI voices to .ogg if we really want to go there. But i still like the idea of variety of voice personalities the AI can chose from.
Of course, I merely brought it up as a fairly popular addition to the variety of options. It would fit well with one of my AI characters while clashing horribly with others (which would be kinda the point I suppose.)
While we're on the subject, how does vox work exactly? Is it just a collection of words (and letters I assume) with pointers to matching ogg files? If that's the case I could probbaly sitch together something like a glitched/stuttering AI voice.
Something like Murphy's voice from Robocop (the very mechanical and stiff style) would be nice too. Definitely would want to make that one from scratch though.
You may have an issue with file sizes and such if you create a file for every single vox word in several voices. Now, this would be AMAZING if you had a TTS engine, but if you are making new files for each word, I predict that you will have problems trying to get it merged as you would make the folder five times larger. Do not get me wrong, I want this, but it is something you really want to get planned out in advance with the maintainers.
if you're wanting different voices, they're going to have to be small lower quality files if you want that merged. However if you want to produce a new standard voice that we can use to make new announcements and such, you'll have more room for file size and quality.
Remie Richards wrote:if you're wanting different voices, they're going to have to be small lower quality files if you want that merged. However if you want to produce a new standard voice that we can use to make new announcements and such, you'll have more room for file size and quality.
Really, I kinda wish we could use this as an opportunity to reshuffle the words in the vox system. There's a lot that literally never get used and a few words that AIs always want to use but never have. I can't really produce an exhaustive list, but if the reshuffle produces less required words per set that would free up budget for more sets/better quality alts
Remie Richards wrote:if you're wanting different voices, they're going to have to be small lower quality files if you want that merged. However if you want to produce a new standard voice that we can use to make new announcements and such, you'll have more room for file size and quality.
Really, I kinda wish we could use this as an opportunity to reshuffle the words in the vox system. There's a lot that literally never get used and a few words that AIs always want to use but never have. I can't really produce an exhaustive list, but if the reshuffle produces less required words per set that would free up budget for more sets/better quality alts
We can prune the least used ones yes, that would free up space, but not that much.
Remie Richards wrote:if you're wanting different voices, they're going to have to be small lower quality files if you want that merged. However if you want to produce a new standard voice that we can use to make new announcements and such, you'll have more room for file size and quality.
Really, I kinda wish we could use this as an opportunity to reshuffle the words in the vox system. There's a lot that literally never get used and a few words that AIs always want to use but never have. I can't really produce an exhaustive list, but if the reshuffle produces less required words per set that would free up budget for more sets/better quality alts
We can prune the least used ones yes, that would free up space, but not that much.
Just for reference, the game is about 110MB and the vox_fem sounds folder alone is about 5MB. Vox files are 1/22 of the entire game. Regardless, I would love more vox voices.
Cuboos wrote:i can convert the Half-life 1 AI voices to .ogg if we really want to go there. But i still like the idea of variety of voice personalities the AI can chose from.
That would be illegal, and any further discussion on this matter will be deleted.
I code for the code project and moderate the code sections of the forums.
Cuboos wrote:i can convert the Half-life 1 AI voices to .ogg if we really want to go there. But i still like the idea of variety of voice personalities the AI can chose from.
That would be illegal, and any further discussion on this matter will be deleted.
Literally the only way you'd be "blatantly ripping-off" the Mesa sounds to point of liability would be to just lift the files from the HL1 directory and plop them into a SS13 directory.
So we're in agreement that ripping the HL1 vox sounds out of the HL1 directory isn't ok.
If you want to make new versions that sound similar/practically the same, that's perfectly fine.
I code for the code project and moderate the code sections of the forums.
MisterPerson wrote:Ok, let me pull up the exact statement you made:
Literally the only way you'd be "blatantly ripping-off" the Mesa sounds to point of liability would be to just lift the files from the HL1 directory and plop them into a SS13 directory.
So we're in agreement that ripping the HL1 vox sounds out of the HL1 directory isn't ok.
If you want to make new versions that sound similar/practically the same, that's perfectly fine.
Which is what I assumed he meant in the first place. Its like you're trying to disagree for the sake of it. Hell, you could alphabet soup the original HL1 sound files and still be free of liability. The only way you'd even be threatened is (as I said) by lifting the files wholesale and dumping them into a directory uncredited. Copyright isn't really as spooky as it sounds.
Generally so long as you don't directly rip content from a game's files, such as noted above, then there's no real concern about a copyright violation. Essentially if you make/sprite/code/voice your own content, even if its based off of something else, you generally should be fine. Just give it a unique name and not "HALF LIFE VOICE".
Copyright is not a good argument, now having unique content is a good argument.
Played around a bit, how would this sound for an AI personality?
It is too grating and difficult to understand. It should sound machine-like, but not overtly so. Think Siri, The HL voice, GLaDOS, or even the current automatic announcement voice.
Hornygranny wrote:It's not your codebase. It's our codebase. You can imply soft power as much as you want, but you don't have it. Division between the server and project is absolute. I'm not interested in reading dezzmont platitudes for the billionth time and won't be checking back in this thread.
Spoiler:
~Simplified for the sake of Wyzack's delicate feelings~ Fuck anti-roleplay suggestions and fuck Bay.
Played around a bit, how would this sound for an AI personality?
Sounds like the AI is a 70 year old ex smoker with a hole in its throat.
I like the voice from the"Wow, this is just like one of my Japanese animes!" (Specifically, animes.ogg). It's clearly not human, but has a bit of emotion to it.
I have the amusing mental image of there being a whole bunch of choices for soundbanks, simply mechanised voices of a big ol' library of prechosen words recorded in various styles.
Imagine a Vincent Price AI.
Hornygranny wrote:It's not your codebase. It's our codebase. You can imply soft power as much as you want, but you don't have it. Division between the server and project is absolute. I'm not interested in reading dezzmont platitudes for the billionth time and won't be checking back in this thread.
Spoiler:
~Simplified for the sake of Wyzack's delicate feelings~ Fuck anti-roleplay suggestions and fuck Bay.
Pybro wrote:
I like the voice from the"Wow, this is just like one of my Japanese animes!" (Specifically, animes.ogg). It's clearly not human, but has a bit of emotion to it.
Everybody likes that voice, it's the standard for all our AI announcements, the issue is all the details on how to recreate that voice have been lost for quite a long time.