Need help with OTOing 'long consonents'

Reelok

Momo's Minion
So this is my first VC voicebank and i really need help OTOing it
ive binged god knows how many hours worth of tutorials on how to OTO
ive managed to do short consonents really well, but long ones like 'shi' and 'tsu' and really just not it

i tried to oto 'shi' but whenever i stretch the pink bar over anywhere, it sounds distorted at the beginning
ive followed countless tutorials so ig its best to just ask here
heres a file of what i tried to do, and ive trialed and errored a lot too, but it just aint working for me

is it something to do wit hthe recording itself or do i just not know how to OTO?

2021-01-13 (2).png
 

WinterdrivE

Ritsu's Renegades
Defender of Defoko
Where you currently have the overlap (green line) is where the left offset (blue area) should be. The red line and pink area look fine. The right cutoff (the right blue area) could stand to be further right since there looks to be some more stable vowel that could be used.

That said, none of this is stuff that should cause distortion or noise. (In fact, I think it's pretty minor and you did a pretty decent job for a first attempt, at least on this syllable) Do you have an audio example you can post?

PS, "ts" is not oto'ed the same way as "sh". It's often easier to think of consonants as "hard" or "soft", where hard consonants are stops and affricatives (in japanese, these are k, g, p, b, t, d (stops) and ch, j, ts (affricatives)) and soft consonants are everything else. This is because stops and affricatives involve a total blockage of air flow whereas the others don't. This means there (in theory) will be a silent (or almost silent) gap before the consonant "starts" (or more accurately, before it's released), so you want to include some space before the consonant and place the overlap before the release of the consonant to simulate this gap. With soft consonants you want to exclude the silence and fade-in before the consonant to simulate the fact that there is no total blockage of air flow, and instead set the overlap across the beginning of the consonant itself.
 
Last edited:
  • Like
Reactions: Reelok and Nohkara

Reelok

Momo's Minion
Thread starter
Where you currently have the overlap (green line) is where the left offset (blue area) should be. The red line and pink area look fine. The right cutoff (the right blue area) could stand to be further right since there looks to be some more stable vowel that could be used.

That said, none of this is stuff that should cause distortion or noise. (In fact, I think it's pretty minor and you did a pretty decent job for a first attempt, at least on this syllable) Do you have an audio example you can post?

PS, "ts" is not oto'ed the same way as "sh". It's often easier to think of consonants as "hard" or "soft", where hard consonants are stops and affricatives (in japanese, these are k, g, p, b, t, d (stops) and ch, j, ts (affricatives)) and soft consonants are everything else. This is because stops and affricatives involve a total blockage of air flow whereas the others don't. This means there (in theory) will be a silent (or almost silent) gap before the consonant "starts" (or more accurately, before it's released), so you want to include some space before the consonant and place the overlap before the release of the consonant to simulate this gap. With soft consonants you want to exclude the silence and fade-in before the consonant to simulate the fact that there is no total blockage of air flow, and instead set the overlap across the beginning of the consonant itself.

Ty for this! i realised after a few tries that tsu isnt the same as shi, but there's something else i need to say,
if i stretch the pink bar any further it causes the distorted sound, but i trialed and errored a bit and finally this is the format that seems to wor kfor my 'shi'
all the tutorials say 'stretch it to where it becomes stable' yet that barely works for any of my otoing. I think it solely depends on how the person has recorded their voicebank, and in my case, i think ive done it slightly differently to everyone else

yeah anyway, i did this for 'shi' and yet it works perfectly fine:
upload_2021-1-15_8-2-36.png

weirdly enough, the 'sh' isnt stretched out too much, but trust, if i stretch the pink bar any further it'll sound whackery-do, and just really distorted at the beginning of 'shi'.
[doublepost=1610698096][/doublepost]btw is there a way to upload audio onto here? (sorry im pretty new to this)
 

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
You can do a screen recording using software like OBS and upload it as an unlisted youtube video.
 
  • Like
Reactions: Reelok

Similar threads