https://uberduck.ai/ logo
Powered by
# machine-learning
  • s
    Just filter out the (screams in pain) (laughs) files from overwatch or your reinhardt will suffer omnic crisis ptsd not uhm... That I would have missed that one..
  • h
    yeah of course, i'll probably end up removing everything in brackets or asterisks (e.g. paladins's disintegration sounds which, exist,)
  • h
    also there's next to no way i'm filtering everything myself so hopefully i'll manage to make an app thing that doesn't import heckign pygame to play audio
  • h
    a web thing maybe
  • s
    Yea I just filtered out everything with brackets. Let me check I might even have the video still on youtube of my best ever rein voice gen fail. Will put in funsynthesis if its still up on yt
  • t
    Got a link to these?
  • h
    well sorta
  • h
    the english ones are transcribed on a wiki https://leagueoflegends.fandom.com/wiki/Warwick/LoL/Audio
  • h
    i already succeeded in reading them off of the paladins wiki so i'll do so for this one too
  • h
    the other-language clips though will have to be transcribed otherwise
  • t
    Just English
  • t
    I'm good on my Japanese audio
  • h
    ah
  • h
    well here's english yeah
  • h
    -oh i just noticed uhh
  • h
    i actually have a thing that converts /mlp/ datasets into ljspeech-style ones
  • t
    Oh you do?
  • h
  • h
    the big script in step 2
  • t
    For this?
  • h
    exactly yes
  • h
    though it ignores the whole phoneme segment
  • h
    although someone suggested it could be used to cut the silence out more reliably than the volume-based approach i use
  • t
    Eh, we'll see
  • t
    I might do another version of this eventually, assuming this is successful
  • t
    So if it comes around then, let me know
  • h
    ?
  • t
    I'm making a machine
  • t
    That clones both English and Japanese voices
  • t
    I have no idea how good it'll be
1...635636637...1068Latest