bshall/hubert

Issues

• Pretraining Hubert on Music

  #17 opened 3 months ago by Bloos
  1

• Finetuning Hubert Discrete and Soft

  #16 opened 7 months ago by slimsushi
  0

• where are the discrete units weights from?

  #12 opened a year ago by thivux
  1

• RuntimeError: stack expects each tensor to be equal size

  #3 opened 2 years ago by rkjoan
  3

• Creating length.json

  #10 opened a year ago by danmathewsrobin
  2

• Can you provide the training script to build the HuBERT-Discrete Content Encoder? Thanks!

  #13 opened 10 months ago by kelvinqin
  1

• "integer division or modulo by zero" error in 158 line train.py

  #15 opened 8 months ago by Lukysoon
  1

• Codes returned by collate function in dataset.py have incorrect shape

  #14 opened 8 months ago by arjunbahuguna
  1

• Confusion about pads and alignment

  #11 opened 9 months ago by splinter21
  0

• why 16kHz ? why not 22kHz ?

  #8 opened a year ago by salisbury-espinosa
  1

• Training accuracy

  #9 opened a year ago by EmreOzkose
  0

• how to train hubert-soft content encoder using Mandarin Chinese dataset?

  #6 opened 2 years ago by joan126
  1

• Error trying to fine tuning the model

  #2 opened a year ago by mathigatti
  1

• your released model

  #7 opened 2 years ago by wyp19930313
  0

• About lengths.json

  #5 opened 2 years ago by EricStuart
  2

• I want to know how to get discrete speech units？

  #4 opened 2 years ago by yaoao2017
  0

• Distributed package doesn't have NCCL built in

  #1 opened 2 years ago by betterftr
  1