https://uberduck.ai/ logo
Join Discord
Powered by
# tacotron-2-support
  • c

    Cris140

    03/21/2023, 8:30 PM
    anything higher will result in Out of Memory
  • c

    Cris140

    03/21/2023, 8:33 PM
    Oh wait, you mean the whisper limit?
  • g

    Gosmokeless28

    03/21/2023, 8:33 PM
    No, Tacotron 2
  • g

    Gosmokeless28

    03/21/2023, 8:34 PM
    I guess I was wrong about that
  • m

    matty1406

    03/21/2023, 8:43 PM
    i always do cutoff 12 seconds anyway in case i wanna give my datasets to 15 when he comes back... that might be the reason why you thought that as well
  • g

    Gosmokeless28

    03/21/2023, 8:54 PM
    Oh yeah
  • k

    k24789304

    03/21/2023, 8:56 PM
    anyways, im still getting the same error after changing those files into 15 seconds
  • k

    k24789304

    03/21/2023, 8:59 PM
    im starting to wonder if it is because i cut them with ffmpeg, lowkey remember ffmpeg doesnt guarantee safe cutting
  • k

    k24789304

    03/21/2023, 9:08 PM
    and now that i think about it maybe i shouldnt cut audios programmatically cuz a sentence might be cut off, trying to somehow solve this problem first tho
  • d

    Dave Johnson

    03/22/2023, 12:46 AM
    Wow, a Site and server mod looks cool, can i be in it please
  • k

    k24789304

    03/22/2023, 7:58 AM
    What is this loop doing? What should be the proper output if I were to print out fname?
    Copy code
    for idx, fname in enumerate(tqdm(filelist)):
        transcription = transcribe(model=model, audio=fname, language = language, beam_size=beam_size)["text"][1:]
        transcript.write(f"{fname}|{transcription}\n")
    I get such an output
    Copy code
    IDX: 0
    Fname: wavs/wavs.zip
  • k

    k24789304

    03/22/2023, 8:33 AM
    cool i think i get it, i skipped the first step to upload the file using this choose files stuff because it was insanely slow so i uploaded myself, and probably in that first step there was unzipping going on and naturally i missed it
  • k

    k24789304

    03/22/2023, 8:39 AM
    is this file upload the best way? wouldnt it be faster to just access the zip file through google drive?
  • k

    k24789304

    03/22/2023, 9:18 AM
    lol yea its hella faster
  • c

    CmanflipYT

    03/22/2023, 10:43 PM
    Idk what I'm doing wrong? I put the dataset on drive just exactly it's at and still not found?
  • g

    Gosmokeless28

    03/22/2023, 10:52 PM
    Is the dataset public?
  • c

    CmanflipYT

    03/22/2023, 10:52 PM
    Yes it is. I always check that if it is first.
  • g

    Gosmokeless28

    03/22/2023, 10:53 PM
    Summoning @hecko
  • h

    hecko

    03/22/2023, 10:53 PM
    uhhhhh
  • h

    hecko

    03/22/2023, 10:53 PM
    oh i see
  • h

    hecko

    03/22/2023, 10:53 PM
    you did MrDrive instead of MyDrive
  • h

    hecko

    03/22/2023, 10:53 PM

    https://tenor.com/bS7DG.gif▾

  • c

    CmanflipYT

    03/22/2023, 10:54 PM
    Ohhh whoops XD
  • c

    CmanflipYT

    03/22/2023, 10:54 PM
    bad spelling moment lol
  • g

    Gosmokeless28

    03/22/2023, 10:54 PM
    Why don't you simply copy the filepath from the folder on the left?
  • t

    TheRoyalRuby2000

    03/22/2023, 10:54 PM
    MRBEEEEEEAAAAAAAAAASSSSSST!!!! BEGIN!!!
  • c

    CmanflipYT

    03/22/2023, 10:55 PM
    I was too lazy
  • t

    TheRoyalRuby2000

    03/22/2023, 10:55 PM
    This is one of the WORST memes I had ever came across!
  • g

    Gosmokeless28

    03/22/2023, 10:55 PM
    I thought copying & pasting the filepath was easier
  • t

    TheRoyalRuby2000

    03/22/2023, 10:56 PM
    It's overused and isn't funny anymore.
1...135136137...158Latest