/tg/Station 13

I saw text to speech being talked about on discord, and I currently LOVE it. But! I have heard of people not playing because of it.
Do you like it? Why? Why not?

Personally, I love how the text to speech is like. I love that I can eavesdrop on people while hiding behind a wall. I like how you can pick a voice for your character now. It makes for fantastic funny moments when the TTS system messes up and says something in a funny way. I love how people are now typing longer (and more spell checked) sentences just so they can hear the TTS more. I love how it incentivises people to talk and interact. Yes there are some kinks that needs worked out (we need more accents from different places for example) , but the fun and enjoyment that it brings me is enough to make me stay up late and play during deadpop hours.
I do understand that I may an odd one out here. So, I would like to see what people think! I put in a poll so I can have a better understanding of what others think about it.

I liked it when it was novel but I’m worried it’s quickly going to feel like a bit of a gimmick. Still a worthy project to try out though, and having some audio indicator that someone is talking to you is something I would still want even if it wasn’t TTS.

Jacquerel wrote: ↑Mon Apr 24, 2023 9:39 am I liked it when it was novel but I’m worried it’s quickly going to feel like a bit of a gimmick. Still a worthy project to try out though, and having some audio indicator that someone is talking to you is something I would still want even if it wasn’t TTS.

I'd love something like 'Don't starve togheter's music chime when a player speaks, or something akin to the different noises produced by goon characters. It should be an option though, so like with tts how you can turn it off.

I think it is kind of neat, but I'm noticing some latency hiccups while it has been active. It could be that I'm particularly susceptible to server load (I am because of my distance from the server), or I might be drawing conclusions from anecdotal evidence. I'd like to see the data once the tests are over to be sure.

The quality of the TTS is good for the most part. Some slightly weird sounding enunciations here and there, but generally pretty okay. I think it helps in some cases, but in a few I think it might stifle emotional conviction because of the neutral tone of the voices. I think the lack of pitch for yelling or whispering can hamper things. It works great for my character, who I imagine to be speaking constantly in that neutral tone anyway, but if I wanted to be more excitable or just emote more, I think it can undermine that.

I think I've noticed some weirdness with whispers being too audible compared to the text as well, which I think ruins some of the potential to misconstrue a whisper for something else because of only getting a fragmented sentence. With TTS, I can just hear the full sentence. I'm sure there can be a solution for that.

TheRex9001 wrote: ↑Mon Apr 24, 2023 11:56 am

Jacquerel wrote: ↑Mon Apr 24, 2023 9:39 am I liked it when it was novel but I’m worried it’s quickly going to feel like a bit of a gimmick. Still a worthy project to try out though, and having some audio indicator that someone is talking to you is something I would still want even if it wasn’t TTS.

I'd love something like 'Don't starve togheter's music chime when a player speaks, or something akin to the different noises produced by goon characters. It should be an option though, so like with tts how you can turn it off.

I was considering making something of the sort when I assumed TTS wasn’t coming back, seems less necessary now although i might still make it for a downstream.

TheRex9001 wrote: ↑Mon Apr 24, 2023 11:56 am I'd love something like 'Don't starve togheter's music chime when a player speaks

I WOULD LOVE THIS AS AN OPTION.

Lot of people see tts from the point of view of it being enabled by default and being used by everyone and thus are disappointed with the current implementation, which makes sense it is lacking in terms of showing emotions and stuff, the first novelai implementation really spoiled us.
However, from an accesibility point of view it's really good and in-depth, so I love it.

when i say hrm it spells it eych arr em. its over.
the tts needs a lotta work still, both in the actual voice stuff so it says words normally, and also needs a lotta tweaks like making nightmares not sound like anime girls and not playing through walls, only where you can see

The voice selection UI could be a bit nicer, as is you have to scroll down the whole list each time to pick the next when you're sampling them.

It's not perfect, but it has potential as is quite fun.

It's nice in some ways and can give life to the words we type, and that's really nice, but there's some pretty big problems with it currently that need ironing out. While latency eating messages is a problem, I don't think it's as big a problem as the other two that I have with TTS. Those being:

1. After some time, the ai seems to bug out and starts every sentence with some complete gibberish phrase, then speaks whatever was written unless it was a short sentence.

2. You can still "hear" voices in other languages, whispered voices, and things that you generally shouldn't. Words and sounds that get replaced by language quirks also come out as "normal". For instance, the chef's "thanks" will still be voiced as "thanks" instead of "grazie" while they're wearing the fake moustache.

I miss the NovelAI api, too bad it's just not really viable

I find TTS to be immersion breaking and the oblivion-tier voice acting to be distracting and annoying, and that's my entirely subjective opinion

What I really don't like is that there are several design issues with its implementation that seem to always be brushed aside:
Why can you hear it through walls?
Why is being able to hear through walls something that is based on a player preference (essentially making it a required choice)?
Why are people allowed to pick their voice when we never even allowed players to pick runechat colors because it broke disguises?

And most importantly why is the entire reasoning given for adding something with this much of an impact allowed to be, quote, "TTS is cool."?

Mickyan wrote: ↑Sat Apr 29, 2023 12:36 am I find TTS to be immersion breaking and the oblivion-tier voice acting to be distracting and annoying, and that's my entirely subjective opinion

What I really don't like is that there are several design issues with its implementation that seem to always be brushed aside:
Why can you hear it through walls?
Why is being able to hear through walls something that is based on a player preference (essentially making it a required choice)?
Why are people allowed to pick their voice when we never even allowed players to pick runechat colors because it broke disguises?

And most importantly why is the entire reasoning given for adding something with this much of an impact allowed to be, quote, "TTS is cool."?

It's in progress; we can get TTS to not go though walls, but having different colored chat is really hard to control.

do you think TG can implement VITS?

Not a fan of the options as I wouldnt play less due to the TTS but I also think it's a step below overrated. The TTS sounds like it's echoing the whole time and doesn't fit well with other sounds. There's also an annoying delay between when the message is sent and when the TTS goes off, so that's a little immersion breaking. Hard to communicate tone over TTS as well, everyone just sounds like they're utterly emotionless. It's kinda funny for a round but until it gets to be more advanced I don't think it adds anything to my experience, and maybe even subtracts from it. A step above Cheridan's april fools meme meme meme meme voice sounds, I'll say that.

Whispering isn't actually silent and funny cult+wizard words need special voice acting done to not sound like trash. I want a deep-bellowing EI-NATH! from my wizard. Doesn't matter if it's just one for male and one for female, we don't need that much variety for the rare stuff, just quality.

i want more voices

we should have Sundowner and Senator Armstrong from Metal Gear Rising: Revengance

The voices also don't work well for lizards. The s sound isn't extended in a lizardlike way, it just sounds bizarre instead.

One of the most incredible additions this game has ever seen

I'm just going to scattershot my thoughts since none of them are particularly deep:

Sometimes it sounds really good, other times it's pretty bad. It struggles with things such as repeat letters, such as someone saying uhhhhhh, or phrases like "y'all"

It's kind of hard to determine if a voice is actually going to sound good in game or not. the "Hello this is my voice" is alright.

I found that it's really bad at giving off tone though. People either sound entirely deadpan, questioning, or they're screaming at the top of their lungs. Nothing can really be done about that though. I've noticed that most of the voices tend to corelate to what a lot of the voice actor's lines are. So, for instance, the Serana voice sounds super deadpan, while the Ellis voice sounds like he's almost always yelling.

It would be nice if the voice menu didn't snap back to the top of the list every time.

Sometimes the TTS is really quick to work. Othertimes, it takes upwards of five seconds to do anything. This sometimes seems to be related to the server's population and/or latency, and other times it seems completely unrelated. I've seen TTS take forever to do anything on a 20 person server with nothing happening and work flawlessly on a 65 person server with half the station in the process of venting into space.

It also doesn't handle people saying sentences really quickly. For instance if you were to say "Yeah I think we got that over here" and follow it with "oh wait, nevermind" then it will just layer both on top of eachother.

No idea if it's still an issue but last I saw you can still hear stuff through walls. I haven't seen any rounds be spoiled by it yet, but one round, I was a cultist and I could hear my cultist friends yelling cult stuff several rooms over. No idea if it got fixed.

Some people have managed to use their voice selection really well for either roleplay or for jokes. A lot of the lizards tend to go with gruffer-sounding voices, or they use the Argonian voice. Felinids use the anime voices most of the time. There was one shift where the HOS had a kid voice, which really brings me back to Gmod DarkRP.

Problems you mentioned should be fixed by now. People sending sentences really quickly will now properly queue the messages up. You can't hear stuff through walls anymore, you'll only hear TTS messages from people who have actually spoken to you (aka you are able to see their chat message).
If you think TTS is taking too long and isn't sending any messages at all, that's probably because TTS is temporarily disabled (potentially because of serverside issues, like TTS being reconfigured, TTS server being down, etc).
You'll be able to tell if this is the case, because no TTS messages will be played at all throughout the entire round

its hard sometimes to play people with social anxiety of lizardism

I find the current iteration to be really, really good, and holding its novelty way past the first rounds I played with it enabled.
I think it's addictive and on its way to become the standard way to play the game. It just breathes so much life into the atmosphere.

A little QOL improvement I'd like to suggest is to make it so browsing and choosing/testing voices is less of a chore.

Right now you can't simply: "next voice>test>next voice>test...", which would be, imo, the most intuitive way for it to work. Instead, the focus resets when you pick a voice, and then you have to scroll all the way to the current selected voice in the dropdown, every single time.

Aside from that, as I said, I find everything great. The variety is great (and the fact more voices keep getting added), the references are great, the emotion the voices show is not perfect but its alright, and at a times very goofy, which I love. It's light years from what was shown a year ago and I'm so glad it's come to this. I personally always play with tts enabled and I'm not planning on disabling it.

everything above, but i think some of the new voices are a bit too recognizable
its a bit jarring hearing the tf2 mercs exactly
and i think the selection menu currently sucks ass

i think the menu could use some work as well

also i am happy with the bebebese option added for those who wish it

It's one of my favorite new features of the game in a while. It really helps to hear a voice and not always have to read a chatbox sometimes. It's also nice for characterization.

My only complaint is very minor in that we don't have a directory of who the voices were modeled from. That's about it.

I have no idea why people are saying it is good, in my experience it messes up every other sentence.

I still like it but it doesn't handle the stuttering from the social anxiety quirk all that well.

this may unironically be the best addition to the game in actual years

improves my rounds anytime it's on

My initial opinion of it was positive and has only grown more positive with time. The new mask filters are also really great as well, everyone working on the system has been doing a great job.

My initial opinion of it was positive and has only grown more positive with time. The new mask filters are also really great as well, everyone working on the system has been doing a great job.

the combine vocoders have caused grayshirt tiding incidents to increase by 27 percent because they want to steal the mask and talk funny

It started off a bit negotive, but I cannot live without now.

I love it but I'm a bit eh on it sometimes because it can't handle stutters all that well, but I put it on the animal crossing esque text blips and it seems fine to me. The downsides of setting it to that is you don't get terry dorms 4 style AAAAAA,AAAAAAA,AAAAAAA! yelling.

Never turned it on, same with runechat, nothing will keep my eyes from reading that white chatbox.