HarryVolek/PyTorch_Speaker_Verification

Issues

• Running train_ speech_ embedder.py reports an error during training

  #79 opened 2 years ago by ZZZzhoup
  0

• perm action

  #28 opened 6 years ago by czy97
  3

• I have trained all data , but how to use it ?

  #78 opened 3 years ago by TonyB22
  0

• how can I reproduce the result: EER across 10 epochs: 0.0377 ?

  #76 opened 4 years ago by JiaqingFu
  0

• new speaker

  #69 opened 5 years ago by ooobsidian
  1

• How to train d-vector model for using on diarization with my own data?

  #67 opened 5 years ago by mesut92
  2

• Single inference

  #75 opened 4 years ago by fabianoluzbr
  0

• error in training

  #72 opened 4 years ago by Jason-1998
  2

• about config

  #68 opened 5 years ago by zyc1310517843
  1

• Shuffling wav files in dataloader does not ensure that all the training files are checked in each epoch

  #73 opened 4 years ago by dkatsiros
  2

• Changing the sampling frequency

  #71 opened 4 years ago by neerajkuhike
  0

• The contrast loss

  #70 opened 4 years ago by amadeusuzx
  1

• train_speech_embedder.py:batch_id

  #44 opened 5 years ago by yahcong
  2

• Very slow training on CPU

  #61 opened 5 years ago by chrisspen
  2

• How to use dvector_create.py

  #34 opened 6 years ago by zeyuanchen23
  7

• Change the code to use multi-gpu, but can not speed up the training.

  #65 opened 5 years ago by Naminwang
  2

• Determining time indexes for embeddings?

  #64 opened 5 years ago by chrisspen
  2

• Unable to run dvector_create.py, file does not start with RIFF id

  #63 opened 5 years ago by chrisspen
  1

• What is the duration of audio of each D vector embedding that is created?

  #66 opened 5 years ago by abhilashnayak
  0

• Unable to test

  #62 opened 5 years ago by chrisspen
  0

• Calculating EER

  #49 opened 5 years ago by nidhal1231
  4

• Workflow of the project

  #38 opened 5 years ago by ahsanmemon
  7

• How to handle multiple speakers?

  #60 opened 5 years ago by chrisspen
  0

• How to Use Your Audio Data Set

  #32 opened 6 years ago by zyc1310517843
  2

• Align embeddings

  #59 opened 5 years ago by pravn
  0

• getting 0 shape array in save_spectrogram_tisv (vctk dataset )

  #57 opened 5 years ago by pranshurastogi29
  2

• Frame rate problem

  #48 opened 5 years ago by nidhal1231
  1

• doubts on the config.yaml

  #53 opened 5 years ago by BarCodeReader
  2

• How to create train_sequence.npy, and train_cluster_ids.npy.

  #56 opened 5 years ago by Sadam1195
  12

• Things to check when testing with different DB

  #55 opened 5 years ago by hash2430
  3

• Possible bug report and suggestion on how to fix it.

  #54 opened 5 years ago by hash2430
  0

• Speaker Verification Predict

  #52 opened 5 years ago by Kailegh
  0

• poor performance

  #46 opened 5 years ago by nidhal1231
  1

• No backend error (in librosa.core.load in preprocessing data)

  #50 opened 5 years ago by nidhal1231
  0

• Overfitting

  #45 opened 5 years ago by biswasingh
  1

• GE2E loss can work with cnn

  #51 opened 5 years ago by wrongbattery
  0

• lstm input wrong shape

  #47 opened 5 years ago by cavus700
  2

• error in data_preprocess.py (save_spectrogram_tisv() function)

  #43 opened 5 years ago by nidhal1231
  1

• Correctness of d-vectors

  #35 opened 5 years ago by nidhal1231
  0

• Embedder-net()

  #40 opened 5 years ago by nidhal1231
  1

• Bitrate/Sample Rate Differences

  #39 opened 6 years ago by bitnom
  1

• about the sampling rate

  #42 opened 5 years ago by nkcdy
  1

• align_embeddings()

  #41 opened 5 years ago by nidhal1231
  1

• DataLoader

  #36 opened 5 years ago by nidhal1231
  0

• About Data_loader.py, "utterance = utterance[:, :, 160]", why we use the num 160?

  #37 opened 6 years ago by Bovey0809
  1

• how to print predict result?

  #25 opened 6 years ago by Aurora11111
  0

• Two question on Embeddings

  #31 opened 6 years ago by mazzzystar
  1

• Question about L2 norm of embedding

  #27 opened 6 years ago by seandickert
  2

• License

  #29 opened 6 years ago by BMillsVT
  1

• Excessive memory use due to train.num_workers > 0

  #26 opened 6 years ago by mkunes
  2