Commits

Branch selector

 e2eb77b

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 25, 2022

• Change GitHub location

  wq2012

  committed
  e2eb77b

  Copy full SHA for e2eb77b

• Add a generate_csv.py script

  wq2012

  committed
  6998077

  Copy full SHA for 6998077

Commits on Oct 31, 2022

• Create LICENSE

  wq2012

  authored
  caadc64

  Copy full SHA for caadc64

Commits on Sep 7, 2022

• Fix the bug in multi-channel to mono conversion. Thanks 韭菜的自我修养 for finding the bug!

  wq2012

  committed
  0e5991f

  Copy full SHA for 0e5991f

Commits on Sep 5, 2022

• Add links to Chinese courses

  wq2012

  committed
  d368243

  Copy full SHA for d368243

Commits on May 13, 2022

• Support full sequence inference

  wq2012

  committed
  f1178f6

  Copy full SHA for f1178f6

• Add specaug pretrained model on all LibriSpeech training data

  wq2012

  committed
  d3b5abe

  Copy full SHA for d3b5abe

• add a get_speaker_encoder() factory method

  wq2012

  committed
  c4c47cd

  Copy full SHA for c4c47cd

• Allow training transformer models

  wq2012

  committed
  0573d96

  Copy full SHA for 0573d96

• Rename SpeakerEncoder

  wq2012

  committed
  25122ff

  Copy full SHA for 25122ff

Commits on May 12, 2022

• polish save/load functions

  wq2012

  committed
  c020898

  Copy full SHA for c020898

• move frame aggregation to the encoder

  wq2012

  committed
  b9bb374

  Copy full SHA for b9bb374

• make loss computation faster with batch processing

  wq2012

  committed
  55aac3f

  Copy full SHA for 55aac3f

Commits on May 11, 2022

• refactor specaug config; use more aggressive specaug config; update specaug pretrained model on clean

  wq2012

  committed
  17715d2

  Copy full SHA for 17715d2

• Add model pretrained on all LibriSpeech training data

  wq2012

  committed
  e25175b

  Copy full SHA for e25175b

• Add SpecAug pretrained model

  wq2012

  committed
  2344da7

  Copy full SHA for 2344da7

Commits on May 10, 2022

• update readme

  wq2012

  committed
  604e91d

  Copy full SHA for 604e91d

• add specaug

  wq2012

  committed
  9066122

  Copy full SHA for 9066122

• move data related code to dataset.py

  wq2012

  committed
  4280d58

  Copy full SHA for 4280d58

• Allow train/eval with data other than LibriSpeech, using CSV files

  wq2012

  committed
  f2a1079

  Copy full SHA for f2a1079

• refactor spk_to_utts

  wq2012

  committed
  3902441

  Copy full SHA for 3902441

• support arbitrary number of channels and sample rates

  wq2012

  committed
  5dfba3f

  Copy full SHA for 5dfba3f

• default to frame-mean bi-LSTM model

  wq2012

  committed
  b696c39

  Copy full SHA for b696c39

• Make eval faster with multi-threading

  wq2012

  committed
  a93d8dd

  Copy full SHA for a93d8dd

• Support training with multi-processing feature extraction

  wq2012

  committed
  e67c23f

  Copy full SHA for e67c23f

• support mean aggregation

  wq2012

  committed
  16b2dc4

  Copy full SHA for 16b2dc4

• support bi-LSTM

  wq2012

  committed
  95ea759

  Copy full SHA for 95ea759

Commits on May 9, 2022

• Allow loading GPU model to CPU

  wq2012

  committed
  5dc9688

  Copy full SHA for 5dc9688

• Update README.md

  wq2012

  committed
  8bc95e1

  Copy full SHA for 8bc95e1

• Add pretrained gpu model

  wq2012

  committed
  0dd31ba

  Copy full SHA for 0dd31ba

• print total steps for training

  wq2012

  committed
  d45d9f1

  Copy full SHA for d45d9f1

• Add a pretrained CPU model

  wq2012

  committed
  5e8999b

  Copy full SHA for 5e8999b

Commits on May 8, 2022

• save models more frequently

  wq2012

  committed
  c8b771e

  Copy full SHA for c8b771e

• fix GPU training

  wq2012

  committed
  f01bd32

  Copy full SHA for f01bd32

• allow GPU training

  wq2012

  committed
  c6d0615

  Copy full SHA for c6d0615

Pagination

Previous

Next