burchim/EfficientConformer

Issues

• warp_rnnt version

  #29 opened 7 months ago by sst0715
  5

• Is it possible to use the VoxCeleb dataset for both training and evaluation?

  #28 opened 9 months ago by hechengzhi985
  0

• ctcdecode

  #26 opened a year ago by HajarMazaheri
  4

• TypeError: wer() got an unexpected keyword argument 'standardize'

  #27 opened 9 months ago by hechengzhi985
  3

• An error when i try to train a Transducer model.

  #24 opened 2 years ago by ZhengWenrui
  2

• Could you share your id of the "LibriSpeech_bpe_1000.model"

  #22 opened 2 years ago by ZhengWenrui
  3

• How to train a model tokenizer on my own datasets.

  #23 opened 2 years ago by ZhengWenrui
  1

• ngram models

  #11 opened 3 years ago by shaojinding
  3

• How to use multiple Gpus for training？

  #21 opened 2 years ago by jiaranjintianchism
  1

• Possible to decouple the encoder and decoder training

  #19 opened 2 years ago by kafan1986
  1

• IndexError: tuple index out of range

  #20 opened 2 years ago by Silentssss
  2

• The Google Drive Link for Pretrained Model no more works

  #18 opened 2 years ago by harisgulzar1
  2

• Can this architecture support streaming ASR?

  #12 opened 3 years ago by kafan1986
  8

• The loss becomes NaN after some training epochs

  #17 opened 2 years ago by harisgulzar1
  7

• Link to Pretrained Original-Conformer Models

  #16 opened 2 years ago by harisgulzar1
  2

• mean loss inf - batch loss: inf

  #5 opened 3 years ago by debasish-mihup
  9

• Building ngram model from my own corpus

  #15 opened 2 years ago by davyuan
  5

• The LM model expected is at word level or at token level?

  #13 opened 3 years ago by kafan1986
  1

• Could you share your pretrained ngram LM?

  #14 opened 3 years ago by davyuan
  2

• Beam Search Decoding giving poor accuracy compared to Gready Search Decoding

  #9 opened 3 years ago by Gouranga95
  7

• Demo Notebook - failing

  #8 opened 3 years ago by TzurV
  2

• Memory leak during CPU inference

  #7 opened 3 years ago by debasish-mihup
  2

• Performance drop when using Distributed Data Parallel

  #10 opened 3 years ago by NarsimhaChilkuri
  1

• LM model required when beam size greater than 1?

  #6 opened 3 years ago by debasish-mihup
  2

• Doubt for training on custom data

  #4 opened 3 years ago by debasish-mihup
  1

• Questions about variational noise.

  #3 opened 3 years ago by maxwellzh
  2

• For the preprocessing

  #1 opened 3 years ago by cecilia930426
  1

• How to run inference using existing models?

  #2 opened 3 years ago by youssefavx
  7