AI VS Sentence Mixing
ฝัง
- เผยแพร่เมื่อ 11 ม.ค. 2021
- Videos used, in the order of appearance:
Steamed Hams But It's Heavy and The Soldier [SFM] - • Steamed Hams But It's ...
TH-cam Poop Skellington's Revenge - • TH-cam Poop: Skelling...
My little pony friendship is gic Pinkie Pie tlt uolliaC - • My Little Pony - Frien...
Steven Jewniverse - • Steven Jewniverse
Steven Jewniverse 3 Threeven Jewniverse - • Steven Jewniverse 3: T...
My Twitter: / constipatedw - ตลก
![dan vs. out of context [season 1]](http://i.ytimg.com/vi/BPcWdc2JT_8/mqdefault.jpg)
I think the traditional style is more entertaining. But that AI is very impressive
I think it's obvious that A.I. it still needs to reach the degree of imitation ... but looking at how we are going, here in 20 or 30 years, it will be almost impossible to distinguish a real human voice from an imitation of artificial intelligence. Just look at those voices from TF2...
We just have to see in time
Alone they are weak.
Together they are strong
yea i like traditional but 15.ai is more good then traditional way
i think that the person who made the ai is impressive. but people who use it over sentence mixing are just lazy
1:24 “Uh, pyro new, oh fuh, my roast is ruin
But watif... I were to purchase fast fewd and disguise it as my own cooking”
Artificial Intelligence at its finest
the laugh too, i've seen some people use the actual laughs and voice clips even when using the ai, like the "myeah" that soldier does, the classic "poutis", and any laughs, either way the ai still has a way to go.
15.ai can be combined with sentence mixing which possibly make it sound really good or really weird
yeah, i feel like in situations like 0:56 , they could've used the original ''Yeah..'' it conveys more emotion, and it would've sounded more natural than 15.ai's version
@@Jedededededede i agree
@@Jedededededede 6:29 It's already there.
Sentence mixing has much more energy to it, and so is good for traditionally fast-pace memeposts, but 15.AI sounds far more cohesive, calmer, and is easier to replicate, making it good for lots of written dialogue that draws out the joke slowly.
Sentence mixing for shouts, text-to-speech for calmness.
drawn out jokes ain funny tho
sentence mixing - funny
15.ai - next gen text to speech
EDIT: you can combine both
It keeps dying
@@youraveragemorononyoutube4435 Sad, really. If that guy wasn't so uptight about removing the generator from the website, he should've figured out how to apply the updates to his source rather than make us feel impatient with using VoCodes.
Chad: AI.15
Lad: Sentence MIXING
*THAD:* SENTENCE MIXING + AI.15
I couldn't agree with you more, sentence mixing is still king.
15.ai is legendary,
but the sentence mixing process is more tradition and have more subtle humour. 2012 gmod vibes
I think they're both good for diff styles of video and people might even mix the two in the future, like how SFM never completely replaced Gmod, but some creators use one or a mix to create their own unique style.
Classic are classics but AI is very impressive and makes everything easy
REAL
15.avi is best for words that the character has not said, while if a full word is already available, use that word to sound more natural
I've noticed that 15.ai can feel emotionless and dull and other times it can feel like someone is directly talking to you. A problem with sentence mixing is sometimes certain lines don't fit the emotion of the sentence but it doesn't have the problem of dull emotionless lines.
I actually think the MLP one worked really well. Maybe it's just to do with the amount of training you give the AI on a certain character.
I honestly could go either way. 15.ai is easier to understand (because sometimes it's hard to understand what a character is saying in the chopped up traditional way), and can actually sound good at times.while the traditional method can have more emotion and isn't distorted. Honestly, if they could design a way for the ai to be able to select emotions for the tts function and have less distortion, it would be perfect. Or at least close.
Omg, I just realised how hard this thing will hit nsfw va market. Or lewd game industry in general.
Sentence mixing are hilarously funny and ai mixing are accurate for longer dialogues
So using both would better
15 ai is funny when they try to be serious. Like many videos of Demoman finding everyone's dirty secrets and it sounds hilarious when they try to be serious and flip out.
Normal mixing is best when it's absolute stupid nonsense and everyone is insane. Then you can also make those weird noises with them as well.
If only Squidward was in 15.ai, then it would be perfect for the skellington's revenge bit.
So how many of y'all have been trying this DeepMoji thing that 15.ai is abusing right now?
You know, the thing where you first type the sentence you want spoken, then a vertical line and then some sentence by which to emote?
For example:
• 「I'm going to post En-Es-Ef-Doubleyou in General!|Happy happy joy joy」
• 「I'm going to post En-Es-Ef-Doubleyou in General!|Oh boy! I can't wait to bomb some dodongos!」
• 「I'm going to post En-Es-Ef-Doubleyou in General!|Sike, that's the wrong number!」
6:48 Hearing the BGM immediately switch to PAL made my spine physically recoil.
The mixing of both is so great!
But yeah, I do think 15.ai can be absolutely amazing when used correctly, but it definitely can't replace sentence mixing entirely
While fifteen.ai doesn’t sync up as well as traditional sentence mixing can (unless it in itself is edited), it definitely sounds more natural and smooth.
But I definitely agree: Combine both traditional sentence mixing and fifteen.ai or vocoder to grow the best dialogue. It depends on the tone and delivery you are aiming for.
@@SmokeTheHorsehog 15.ai strong soot is one word or a long sentence
Flame Mane --- except for Heavy, for some reason it sounds like he's a grandpa or something like that.
For shitpost i prefer sentence mixing
But for projects where I require more natural voices, I prefer 15.ai
Now if only 15.ai stayed up for more than a fucking week at a time :o)
If only there was an outdated version of the site that receives far fewer updates but stays up for longer periods of time... HMMMM.
I don't think the creator of 15.ai has heard of mirrors.
@@flashn00b Yup, and I’ve talked about that on Twitter too, most of 15.AI’s fans seem like condescending teenagers (admittedly I am a teen myself but still, if they are older they are likely condescending) and kids, both talking about how the guy running 15.ai works “Soooooooooooo hard” for his fans and that I have no excuse to bash him for his shitty updating methods
@@satan2583 He offers his service for free, he works on it alone, and most importantly, it's a study project. Allowing people to entertain themselves with the AI is literally a bonus, so we should be lucky he decided to share the tool at all.
@@Shyfire If I offer a normal, seemingly constant service, and then periodically take it away with little explanation besides “improvements” you’d likely get pissed too.
@@satan2583 I'd be disappointed, but not pissed. The guy owns nothing and never promised anything. He's making this whole thing for a study project, and for all we know he might even stop working on it once it's done, especially if he only gets negative feedback because updates dont go fast enough.
I think the sentence mixing is a bit better because it has a certain charm to it.
Sentence mixing sounds better and makes for much more creative dialogue, but is way harder to do.
I think a good mixture of both will work the best in most vids
15.ai is too calm and Sentence Mixing can't sound natural sometimes.
Use both.
that spongebob ytp gave me nostalgia I didn't know I had
5.ai really depends on how many voice samples there are of the character. Like the SpongeBob one sounds amazing, probably because there are a bajillion episodes for the ai to learn from.
Sentence Mixing is still but with both i think it good ideal
Sentence Mixing sounds weird, but does the job right.
AI does better grammar-wise, but sounds more lifeless sometimes without tweaking.
Both have their advantages.
I prefer the traditional style as its funnier easier to display emotion without sounding robotic.
I feel that using both is the best option.
Mixing AI and sentence mixing might result in hahahahaha funny
I think if we mix them it’ll be perfect
In the year 2077 voice mixing and ai becomes one making them
Cyborg
15.ai: used for generic sentences and versatile speech
sentence mixing: funnier, more fluid, and makes for the better ytp
Yea 15.ai is cool and all, but nothing can replace the charm and originality of sentence mixing
I like traditional because It seems like there’s more emotion
Why not both?
Sentence mixing is more ethical.
4:33
“𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵𝗵”
Both of them are as good as each other, but I think it comes down to how you use them.
Sentence mixing is best used in SpongeBob
Chad mixer vs spongbob ai
Ho ho ho, that swipes me on the knee
What I would is combine the both and use the AI when the sentence Mixing is too distorted to understand
A mix of sentence mixing and ai works best
Sentence mixed is more energetic
15ai however is more understandable
i like things made by humans
I remember watching that spongebob ytp a long time ago. I'm glad i'm not the only one to remember.
Examples of characters on 15.ai
TF2 classes
MLP characters
Crystal Gems
Persona characters
Spongebob
The Smash 4 announcer(formerly)
Dude I love the angry birds theme
I think the most effective method - although it would take more time - would be to mix the two, primarily in doing the lines first in 15.ai and then correcting the egregious mistakes of the AI manually, either with combining multiple 15.ai interpretations or with manual sentence mixing.
I feel like 15.ai needs to work on emotion since sentence mixing can either sound great or wrong emotions sometimes. But for general talk like round 2 that doesn’t really need any emotions that’s where 15.ai shines. Also I feel like 15.ai is better when your tryna go for a beat like talk. What i mean is when your tryna make a character song a specific tune and sometimes sentence mixing doesn’t help. Sentence mixing sounds more funnier then 15.ai as it comes out more fresh. When trying to make a funny phrase or word 15.ai makes it sound weird, more general and sometimes wonky and sounds like a voice crack. 15.ai is easier but that doesn’t mean it’s as efficient while in the other hand sentence mixing can be hard but it’s worth the time. I’m
more into sentence mixing but 15.ai can sometimes shine. Also I forgot laughing or spamming a letter like “hhhhhhhh” it’s blursed.
The exaggerated head movements seem so out of place with the AI voices. Also, "OH FAAAHK!" > "oh, fuck."
The AI has some potential, but it’s still very limited. The best results are shot 1 sentence lines. Anything more than that begins to sound unconvincing and monotone.
It’s best to use the AI in conjunction with sentence mixing or or with actual grunts and breathing of the characters of clean rips of them are available. You could probably make decent Visual Novel or comic dub.
The thumbnail is very funny
You can barely understand the original version. I love you AI
Of course ytp's were used as examples.
I Like Both
The TTS ai is very impressive, but the drawback is that you can tell when it cheats. There’s a certain tinny quality when it’s splicing tones together
Gonna have to give it to sentence mixing.
i think, this ai is a future of gmod and sfm. But if we talk about casual poops or tv show - traditional style is better, cuz it send to us emotions/
15 ai is better in comics
Sentence mixing is funnier
“Pyroh noooh. Oh fuah.”
The AI makes it sound kinda weird
I prefer sentence mixing when used in comedic animations, but 15.ai is better with more serious styles
Sentence mixing and 15 A I are pretty good in my opinion 15 AI being used to voice the mythical characters is pretty legendary.
You can fine tune the 15.ai so that it has emotion and isn't so artificial if you really try and spend a long time with it
15.ai just makes the character have no emotion, but it can help create words that are hard to sentence mix.
There's instructions in the website on how to get the ai to be more emotional and pronounce certain words.
ai is cleaner and more understandable
but sentence mixing is way better and way funnier
Use both, for words that is harder to use while mixing fill it in with 15.a
lol the ai is pretty good but almost sounds demonic
It's a bit harder to understand traditional but funny since it's the actual voice acting, 15.ai sounds robotic but understandable and I imagine it will only improve in making it sound believable in the future...I think traditional really is funnier though.
There any other apps like 15.ai? Because its been down for months
The last sound of round 3 is so fucking creepy 4:32
Sentence mixing is better for comedy
I get why they pin them against each other but AI with be the future with sentence mixing you just use ai for syllables that you need
There made for each other
Ai is great for small sections but cant make full sentences
Sentence mixing is better overall but your limited to what you have
Ai can also help with context
Ai sounds a bit like they smoked alot
Honestly I was impressed by 15.ai in every regard except for tf2.
I say use both .
fuck man i miss mobrosstudios
What is the opening song from?
Honestly sentence mixing is way funnier.
Both sound good, but sentence mixing sounds funnier IMO.
Krobo really is the only one who unarguably beats 15.ai
SPONGEBOB SOUNDS WAYYY BETTER THAN THE TF2 THING
I think that the best way to compare the two is with TF2 voices and videos, since unlike Spongebob, MLP and Steven Universe, there isn't as much context to TF2 related creations.
Look at the examples brought in the video itself: The MLP poop took place in existing scenes with entire lines of dialogue staying nearly uncut, so of course the original voice acting is superior to AI generated lines, and the SU poop was spoofing two *songs*, something the site can't do at all, and using it as a point of comparison between sentence mixing and AI is kind of pointless because the whole point of that poop was to screw around with the songs, their timings and tones.
Then compare that with the TF2 video, where aside from a few lines uncut from original game dialogue, most of these lines were pretty much hand made using sentence mixing, and this is where the true difference between SM and AI really shows, because creating completely new lines takes a TON of work with sentence mixing (something you can see on videos like Heavy Is Dead and a few highly edited GMod videos), while using 15.ai everything can sound much smoother and you can still infuse your own tone into the sentence using some specific coding for it (a great example for 15.ai dialogue done well is with Spy's Confession which sounds almost like it was acted by the original VA's.)
In the end I think that you can essentially use both if you want to get something really good, but AI has a ton of potential left to be explored and polished and I think that if people want to have custom sentences that sound smooth and believable it's pretty much the best way you can go about it.
(I will still admit that good sentence mixing hits much different and can never be replaced in my heart)
The AI is always very scratchy sounding and it usually cant capture emotion correctly. Although it is really cool, I prefer sentence mixing.
15.ai has no emotion to it
That's not true. It is possible to put emotions using some commands. Besides, emotions are generated automatically by an AI that adds emotion into speech.
@@eljote3 oh ok
I think they both are good but 15.ai is for more seriousd sfuff while sentence mixing is fir silly/ not serious things
I'm not a native English speaker but i could barely understand the sentence mixing one
While the AI is certainly impressive, it just doesn't have the same charm or personality as sentence mixing.
PvZ music made this better.
Sentence mixing is funnier.
Trade off of replacing sentence mixing
-Less emotion
-More likely for Clear sentance mixing
-not as entertaining
15.ai just give me creeps when it comes to TF2. I literally get uncomfortable listening to it :(
traditional sounds a lot better but is a lot harder to understand. and 15.ai is kinda bad sounding and is kinda lifeless. but it does the job.
I don't like 15.AI it sounds weird robotic
It is impressive but its weird i dont like it.
Sentence mixing is the winner, and Sentences mixing is way funnier. 15 ai is the loser, and 15 ai is not a good source and it wasn't so funny cuz its sounds worst as robotics, alienish, sex, posession, sore throats, unwell, unclean. And it has got to go away again and never return.