https://uberduck.ai/ logo
Join Discord
Powered by
# machine-learning
  • h

    hecko

    12/30/2022, 6:13 PM
    also alignment graph ≈ attention graph really
  • h

    hecko

    12/30/2022, 6:13 PM
    same thing
  • h

    hecko

    12/30/2022, 6:13 PM
    https://github.com/keithito/tacotron/issues/144
  • m

    mepc36

    12/30/2022, 6:23 PM
    ooo didn't know that, thank you
  • m

    mepc36

    12/30/2022, 10:29 PM
    Any idea what package carykh is ysing in this video to identify the first and last phonemes that he needs to quantize to the beat (at 11:37 of the video

    https://youtu.be/a0EyfdQ0QTQ?t=697▾

    )?
  • h

    hecko

    12/30/2022, 10:30 PM
    he's just doing it based on amplitude
  • h

    hecko

    12/30/2022, 10:30 PM
    in other videos though he used
    gentle
    for phoneme alignment
  • m

    mepc36

    12/30/2022, 10:30 PM
    like manually using ableton or cakewalk?
  • h

    hecko

    12/30/2022, 10:30 PM
    nnno
  • m

    mepc36

    12/30/2022, 10:31 PM
    what tools he using to identify the amplitude that's greater than 0.2? sorry i phrased it poorly
  • h

    hecko

    12/30/2022, 10:31 PM
    no clue
  • h

    hecko

    12/30/2022, 10:31 PM
    but there are many
  • h

    hecko

    12/30/2022, 10:31 PM
    what programming language do you like
  • m

    mepc36

    12/30/2022, 10:33 PM
    node.js/javascript
  • m

    mepc36

    12/30/2022, 10:33 PM
    or python if thats all there is
  • h

    hecko

    12/30/2022, 10:33 PM
    i don't know much about javascript
  • h

    hecko

    12/30/2022, 10:33 PM
    but
  • h

    hecko

    12/30/2022, 10:36 PM
    ...hm i can't find any decent libraries
  • h

    hecko

    12/30/2022, 10:36 PM
    in python i'd use pydub
  • h

    hecko

    12/30/2022, 10:36 PM
    but this is simple enough that you might be able to just operate on the audio as an array
  • m

    mepc36

    12/30/2022, 10:41 PM
    hmmm good point let me try that
  • m

    mepc36

    12/30/2022, 11:30 PM
    k i think i found some...I might I be able to do the same quantize-first-and-last-syllable approach as carykh using one of the peak detection pkgs listed here: https://github.com/MonsieurV/py-findpeaks
  • m

    mepc36

    12/31/2022, 12:43 PM
    I appreciate your help yesterday man, you got a patreon/buy a coffee/etc. that I could sign up for?
  • h

    hecko

    12/31/2022, 12:43 PM
    i do happen to have a liberapay
  • h

    hecko

    12/31/2022, 12:44 PM
    but ehh i don't think that was worth a money
  • t

    tanooki426

    12/31/2022, 5:47 PM
    Can someone please explain what this means? I'm trying to check if my WAVs are valid
  • t

    TheRoyalRuby2000

    12/31/2022, 5:51 PM
    No idea why that’s happening.
  • h

    hecko

    12/31/2022, 5:57 PM
    means colab decided to be evil and make an invisible directory because you dared to use the file explorer
  • h

    hecko

    12/31/2022, 5:57 PM
    make a new cell, type
    !rm
    , space, then paste in the path it gave you
  • h

    hecko

    12/31/2022, 5:57 PM
    run it, then run this agagin
1...102910301031...1068Latest