https://uberduck.ai/ logo
Powered by
# machine-learning
  • a
    ok
  • a
    and should Mr. Bianco be changed to Mister Bianco?
  • a
    alright @User i see what you're saying
  • h
    it automatically does that but for clarity i'd do that yeah
  • a
    so i need to spell everything out ok
  • k
    @User , after training this I cannot seem to get meaningful text. Its a wierd buzz sound
  • h
    are you using the hifi-gan notebook
  • h
    if so disable 
    pronunciation_dictionary
  • h
    because we train directly on text, while the default synthesis settings are for phoneme-trained models
  • a
    Hi @User when you get a chance here is the revised version. Is there anything else I should do to correct it?
  • k
    Alright thanks. I now get a voice. Does the character matter ?
  • h
    you didn't change the numbers which are the main concern
  • h
    also remove the blank line at the end
  • h
    character as in?
  • h
    oh no that doesn't matter
  • h
    those are the preset voices, when you paste a model id into 
    tacotron_override
    it ignores that
  • k
    Alright thanks, am gonna come for more lessons I guess. I want to get a hang of this
  • a
    I hope this is the updated version
  • a
    @User can you plz check one more time and make sure im good to load it in the tacotron?
  • a
    @User I got disconnected from the Uberduck Colaboratory, so I am now back to running step 2.
  • h
    i see the number 
    650
    , if that's pronounced "six hundred and fifty" then no problem but if "six fifty" then change that
  • h
    also 59.wav is missing, is that intentional or
  • a
    sorta intentional, didn't like that one
  • a
    Also I don't see 650 in my list
  • a
    in my new one
  • h
    152.wav
  • a
    yep
  • a
    fixed
  • a
    also i got disconnected from colaboratory, so is it alright if we start from square 1?
  • h
    yeah
1...431432433...1068Latest