google/uis-rnn

Issues

• assign gpu with arguments

  #91 opened 2 years ago by pthavarasa
  0

• Predicted labels doesn't match with Ground truth labels but the accuracy of test results is 0.8%

  #90 opened 2 years ago by SanaullahOfficial
  1

• [Bug] Making a prediction on CPU after training on GPU

  #89 opened 2 years ago by pthavarasa
  0

• Any documentations on training from scratch using custom data in other languages ?

  #88 opened 2 years ago by thangld201
  1

• rnn initial state trainable

  #87 opened 3 years ago by jardnzm
  1

• uis-rnn can't work for long utterances dataset?

  #50 opened 5 years ago by wrongbattery
  19

• Is there a way to fine tune an already existing pre-trained model?

  #86 opened 3 years ago by saumyaborwankar
  1

• [Bug] Predict method does not finish

  #73 opened 4 years ago by ArlindKadra
  3

• [Question] About num_non_zero

  #84 opened 3 years ago by shubhunegi30
  1

• [Question] The dimension of toy test data [test_sequence] is (25, 95, 256) what does the first 2 dimension represent? Toy train data [train_sequence] has dimension (4627, 256) which is understandable.

  #85 opened 3 years ago by saumyaborwankar
  1

• Is is possible to pre-load the model for multiple request?

  #83 opened 3 years ago by zhixun-luy
  1

• No module named coverage

  #82 opened 3 years ago by yerzhan7orazayev
  1

• Change input size

  #81 opened 4 years ago by surasakBoonkla
  1

• [Question] Are input d-vectors for training assumed L2-normalized?

  #80 opened 4 years ago by vadimkantorov
  8

• Unable to convert pytorch model to tensorflow in Diarization on mobile device.

  #79 opened 4 years ago by alamnasim
  2

• how to control the number of different speaker when predicting?

  #78 opened 4 years ago by NewEricWang
  1

• uis-rnn gives different result on broken audios and continuous audios

  #76 opened 4 years ago by ashu170292
  5

• Question about custom data generator

  #75 opened 4 years ago by YanaHontarenko
  0

• what is train data format?

  #74 opened 4 years ago by jkkj1630
  1

• [Question]About UIS-RNN d-vector

  #67 opened 5 years ago by BarCodeReader
  7

• How to create training data?

  #62 opened 5 years ago by chrisspen
  2

• about model

  #72 opened 4 years ago by zyc1310517843
  1

• Embedding Extraction Procedure

  #71 opened 4 years ago by divyeshrajpura4114
  1

• I run the demo use toy data, found loss is negative number.

  #69 opened 5 years ago by Naminwang
  1

• Confusion about predicted labels

  #68 opened 5 years ago by clabornd
  3

• Any plan on upgrading to pytorch 1.0+?

  #47 opened 5 years ago by tanvirfuad
  3

• [Bug] Incorrect estimation of transition_bias

  #55 opened 5 years ago by DonkeyShot21
  13

• Repo

  #66 opened 5 years ago by chrisspen
  1

• Handling unknown speakers?

  #64 opened 5 years ago by chrisspen
  3

• Handling unknown speakers?

  #65 opened 5 years ago by chrisspen
  1

• uis-rnn error is too high.

  #63 opened 5 years ago by wychoi44
  1

• [Question] Online Speaker Decentralize

  #61 opened 5 years ago by hungnvk54
  0

• question about the sigma2 prior loss increasing during the training

  #60 opened 5 years ago by maxandchen
  2

• [Question] How can I get the text of each speaker after speaker diarization?

  #59 opened 5 years ago by rowhitswami
  3

• Loss calculation in prediction

  #58 opened 5 years ago by gen35
  2

• Is UIS-RNN the technique used in Speaker Diarization in Google Speech-to-text API? [Question]

  #54 opened 5 years ago by blazejdolicki
  1

• Understanding the use of resize_sequence() and the batch creation for RNN training

  #53 opened 5 years ago by vickianand
  2

• [Question] how to relate sigma2_prior_loss() with the paper

  #52 opened 5 years ago by jprobichaud
  1

• Understanding diarization labels

  #51 opened 5 years ago by zyc1310517843
  6

• [Question] Which feature was used for VAD?

  #48 opened 5 years ago by seungwonpark
  1

• The number of speakers and whether the content of the speaker needs to be the same。

  #49 opened 5 years ago by zyc1310517843
  1

• How to convert audio data into test data of algorithm for testing

  #46 opened 5 years ago by zyc1310517843
  1

• [Question] Can I use speaker annotated datasets in other language rather than English?

  #44 opened 5 years ago by muntasir2000
  1

• How to define a speaker per segments from overlapped widow of frames? Prediction on realtime data?

  #43 opened 5 years ago by alamnasim
  5

• The clustering performance influenced by overlap window size

  #41 opened 5 years ago by taylorlu
  7

• Is the GRU really needed to predict mu_t ?

  #42 opened 5 years ago by hbredin
  7

• [Question] How to prepare embedding data for training UIS-RNN?

  #37 opened 5 years ago by OpenCVnoob
  5

• [Question] Would it be possible to publish trained models

  #39 opened 5 years ago by rnunziata
  3

• [Question]test on speakers untrained

  #40 opened 5 years ago by Aurora11111
  3

• [Question]Performance degrade for different win size

  #38 opened 5 years ago by FengLeee
  1