wespeaker

wespeaker copied to clipboard

About implement of Normalized Maximum Eigengap Spectral Clustering(NME-SC) for Speaker Diarizaton

2

Thank you for uploading pre-trained ECAPA-TDNN model. For speaker diarization, the spectral clustering algorithm used by wespeaker uses the p-neighbor binarization scheme, and "p" should be choosed by people. I...

Zhubisong

pre-trained models trained on only voxceleb1

1

I'm benchmarking speaker embedding models, for speaker verifications, that are trained and tested on the [voxceleb1 dataset](https://www.robots.ox.ac.uk/~vgg/data/voxceleb/vox1.html). I am referring to the pre-trained models list [here](https://github.com/wenet-e2e/wespeaker/blob/master/docs/pretrained.md) but it looks like...

sidhantls

Train on new Language

9

Hi, there is a recipe to follow on how to train on a new Language? Thanks a lot

emanueleielo

When will Android platform deployment be supported?

8

Thanks for this repo.Excellent recognition results！ Looking forward to open source code for Android platform deployment.

tianmala

QMF and fusion in VoxCeleb Speaker Recognition Challenge 2023 papers

7

greatly thanks to the project of speaker verification. in VoxSRC 2023, many team use QMF and ASnorm to improve the score, seems QMF is as good as ASnorm, if it...

wendongj

Support for Calculating Loss and Accuracy on Validation Data?

2

I'm currently in the process of training a model and have been tracking the loss and accuracy metrics. However, I've noticed that while I can calculate these metrics for the...

fukudatppei

fine tuning the pretrained model

3

How can i fine tune the pretrained model with my own audio files for speaker verification? There is no particular tutorial for it and i'm lost

Emo3032

It would be nice if you could `extract_embedding `with a `batch_size` parameter. Also accepting Numpy arrays or torch tensor instead of only file like types would be nice.

asusdisciple

Hi, I know there is [a plan to add 3D-Speaker](https://github.com/wenet-e2e/wespeaker/blob/efe6df2f0c6a5f76b3fa9092381d340a7c1e9a9b/ROADMAP.md?plain=1#L18). Is there any date for this issue? Are you planning to release pre-trained models on 3D-Speaker ?

EmreOzkose

Is it possible to implement dataloader for triplet loss/GE2E

7

Hi, is it possible to make a batch containing M speaker and N utterances for each speaker?

mmmmayi