search

jzi040941 / PercepNet

Unofficial implementation of PercepNet: A Perceptually-Motivated Approach for Low-Complexity, Real-Time Enhancement of Fullband Speech
BSD 3-Clause "New" or "Revised" License
332 stars 92 forks source link

Data Creater(c++) for pytorch #4

Closed JiachuanDENG closed 3 years ago

JiachuanDENG commented 3 years ago

Hi jzi040941! Thanks for sharing your code with us! But may I ask where can I find the code for Data Creater(c++) for pytorch?

jzi040941 commented 3 years ago

hello JiachuanDENG, main.cpp, denoise.cpp are c++ code related to Data creator here I write briefly how to create data for training. (will add more detail information about preparing data and training on README file on this weekend)

first, you need to create main binary file

g++ *.cpp *.h -o main

second, main for example, I use sample pcm file on sampledata folder that I get it from VCTK, DEMAND dataset.

main sampledata/speech/speech.pcm sampledata/noise/noise.pcm 4000 test.output

third, convert this binary file to hdf5 format file using bintoh5test.ipynb (this ipynb is quite dirty now.. if you can improve more feel free to make pull request) based on binto5test.ipynb it will generate h5.training file.

as you know this repo is unofficial and still on progress also side project for me. It might have problem, make issue or create pull request If you find any weird thing or can make improvement

JiachuanDENG commented 3 years ago

hello JiachuanDENG, main.cpp, denoise.cpp are c++ code related to Data creator here I write briefly how to create data for training. (will add more detail information about preparing data and training on README file on this weekend)

first, you need to create main binary file

g++ *.cpp *.h -o main

second, main for example, I use sample pcm file on sampledata folder that I get it from VCTK, DEMAND dataset.

main sampledata/speech/speech.pcm sampledata/noise/noise.pcm 4000 test.output

third, convert this binary file to hdf5 format file using bintoh5test.ipynb (this ipynb is quite dirty now.. if you can improve more feel free to make pull request) based on binto5test.ipynb it will generate h5.training file.

as you know this repo is unofficial and still on progress also side project for me. It might have problem, make issue or create pull request If you find any weird thing or can make improvement

Thanks jzi040941! Let me check it out! Good job!