wngh1187 / exu-net Goto Github PK
View Code? Open in Web Editor NEWPytorch implementation of Extended U-Net for Speaker Verification in Noisy Environments
License: MIT License
Pytorch implementation of Extended U-Net for Speaker Verification in Noisy Environments
License: MIT License
Hello, I recently found a problem when reproducing the experimental results, the original was to use multi-GPU training, but these two days, the resources are limited, and found that the model can not run normally on a single gpu training, may I ask you have encountered this problem, do you have a single gpu training version, thank you very much, here is my error:
Traceback (most recent call last):
File "/home/wangzh22/anaconda3/envs/my_pytorch/lib/python3.9/site-packages/torch/multiprocessing/spawn.py", line 59, in _wrap
fn(i, *args)
File "/home/wangzh22/pycharm/ExU-light/main.py", line 140, in run
trainer.run()
File "/home/wangzh22/pycharm/ExU-light/trainers/train.py", line 31, in run
self.test(epoch)
File "/home/wangzh22/pycharm/ExU-light/trainers/train.py", line 125, in test
self.embeddings = self._enrollment()
File "/home/wangzh22/pycharm/ExU-light/trainers/train.py", line 201, in _enrollment
embedding_dict[keys[i]] = embeddings[i]
TypeError: unhashable type: 'list'
Hello, I saw that ExU-Net-L was mentioned in the article. But it seems that it is not given in the code you gave. Can you tell me what changes have been made based on ExU_Net?
Hi, I trained the model according to your guidance, but is there a special testing section? What if I want to test the trained model on other datasets?Such as Voices ?
In main.py line 177, one parameter 'process_id' for function 'run' is missing?
Hi,
Thank you for making the code & pretrained models public. I am trying to reproduce the results. However, the result I got after running your pretrained ExU-Net model on VoxCeleb1-O is different from the paper's result. My reproduce result is EER_clean: 0.22067868504771956
. To get the result, I just commented the train
step and load the pretrained model in the test
step. I am asking whether you have encountered this problem?
Once again, thank you for sharing the code. I am looking forward to hear from you!
Hello, I'm trying to replicate your results on the VOICES dataset, but I'm having huge problems,how do you use VOICES as an additional verification set, I see that its trails list and voxceleb1 is very different, voxceleb1 is the audio at both ends and then gives 0/1 to represent whether it belongs to the same speaker, but VOICES is a column of labels, a column of audio Then imp/tar, how do you use it, or did you randomly make a list of trails like voxceleb1, I would like to humbly ask you, I have never encountered this kind of list before, and it is the first time to contact the VOICES dataset, I don't know how to deal with it,I see that you have not uploaded the relevant code on github, perhaps it is convenient, can you tell me how to deal with it, thank you very much for your help!The format of the Voxceleb1 and VOICES datasets is as follows:
(1)Voxceleb1_trails:
1 id10270/x6uYqmx31kE/00002.wav id10270/GWXujl-xAVM/00035.wav
0 id10270/x6uYqmx31kE/00002.wav id10306/uzt36PBzT2w/00001.wav
1 id10270/x6uYqmx31kE/00002.wav id10270/GWXujl-xAVM/00038.wav
(2)VOICES_dev-trial-keys.lst:
Lab41-SRI-VOiCES-rm1-none-sp3446-ch144019-sg0006-mc03-stu-mid-dg080 sid_dev/sp3521/Lab41-SRI-VOiCES-rm2-musi-sp3521-ch012715-sg0017-mc04-lav-mid-dg090.wav imp
Lab41-SRI-VOiCES-rm1-none-sp3446-ch144019-sg0006-mc03-stu-mid-dg080 sid_dev/sp3521/Lab41-SRI-VOiCES-rm2-musi-sp3521-ch012715-sg0006-mc10-lav-cec-dg120.wav imp
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.