[Arknights] W English Dub (ElevenLabs Fanmade)

[@Absolute__Zero]

I didn’t think this AI was capable of sampling Japanese voice lines and creating an English voice out of it :O

[@wiks50]

I see you even here, you can't escape me xD

[@Absolute__Zero]

@@wiks50 What are you doing in an Arknights video xD You've been playing the game too?

[@wiks50]

@@Absolute__Zero Perhaps

[@konohamaru77]

Maybe not today...but...in the future...you just have to wait

[@alienelephant4721]

I honestly thought JPN voice actors speaking english would sound different from english voice actors but surprisingly they are quite similar.

[@curious2882]

The fact that you input japanese and it output english this good is insane

[@Eniteee]

the heck. the ai even had the tones and expressions when it talks

[@comdnoive]

It's like British and Southern at the same time.

[@psychlyeslg]

Kinda make sense if she's with brits and southerners.

[@hereticloli2476]

Hot damn this ones nice! Heres to hoping HG hire a great EN va for such a great character.

[@oherman]

This is one of the better renders of her voice. Of course we can't do anything about generating a convincing laugh right now, but this is better than nothing!

[@xxxod]

this ai makes me think that in the future voice actors wont have jobs

[@oherman]

@@xxxod not really. It still has a lot of bumps and things left wanting, especially if we're talking about consistent realism.

[@akin624]

@@oherman With how fast the ai technology is growing, I won't doubt that it'll most probably pose a real threat to not just the voice acting industry, but the entertainment as a whole. I mean this is still the earliest stage of this ai, give it a decade and by then the tech might already be on par or even surpass human work, mostly in efficiency. It's both fascinating and terrifying at the same time if you think about it.

[@SoundGuideline]

She actually sounds more like I expected W to sound before I heard the JP voice.

[@Miku-yu5iu]

thats like magic holy shit

[@hoshiburaito]

it was almost the same as the voice in japanese, wtf this ai is very good

[@adhitya105]

woahhh.. BRIISH ayana

[@KartoshkaMakaroshka]

"in battle 4" and "3-star result" сильно мимо персонажа. А так понравилось, получилось интересно :)

[@polkalamypekopeko2969]

Probably as good as it is gonna get.

[@wenthusiasts2470]

MY QUEEN!!!!

[@aldimahdi9141]

This ia amazing

[@herbert-kenumera960]

Science just dropped a helluva banger.

[@shireii8764]

woah

[@shinning123]

This should be official, better than the jp version with a bit more emotions

[@Wan_Setiawan]

This is Wuh

[@ofal5124]

who can't even protect a single sarkaz damn stingy~

[@rasunin]

how long did you play with the Voice Settings? the 'Stability' and 'Clarity + Similarity Enhancement' sliders.

[@Kulgia]

Depends. Several times I think. I didn't touch Clarity that much. I usually play around with Stability

[@yaek7311]

@@Kulgia would you mind sharing the slider percentages and samples you used?

[@Kulgia]

​@@yaek7311 I used the Japanese voice samples I got from KZbin. I try finding the whole voice lines without any background music. For clarity, is either 75% or more. For this W one, I don't remember what clarity I used but it's constant. Depending on the audio file, if the quality is good I put it on high, and its stays there for the rest of the voice lines. Stability, really depends on your character's personality. For W I put it on lower than 50%, although it goes weird and crazy sometimes, I play around with it until I find what I think is perfect.

[@yaek7311]

@@Kulgia Thank you! It sounds to me like you used one Japanese voice sample per one English line output. Is that correct? Or did you combine all samples you found into one file to clone a universal voice for all English lines?

[@Kulgia]

@@yaek7311 It's the latter. I combine all the samples. I find that easier than doing one by one Japanese sample per English voice output, which I doubt it'll come out any better if doing that.