Train LA Baseline-LFCC-LCNN with different features about 2021

[KyloRen1]

@joaquinCaceres Hi, apologies for the off-topic.
Were you able to set up the conda env? As I am trying to set it up for hours now without any progress due to the package conflicts.

Thanks!

[TonyWangX]

@joaquinCaceres Sorry that I didn't notice the email notification.
Unfortunately, HTK format is not supported right now.
(It can be done, it but requires time to debug).

Maybe a simple way is to copy the LCNN model definition and plug it in your project

2021/LA/Baseline-LFCC-LCNN/project/baseline_LA/model.py

Line 164 in 17c3ce9

self.m_transform.append(

[TonyWangX]

@KyloRen1 Hello, could you tell us your software/hardware environment (OS, kernel, GPU...) and paste the message on conflicting packages?

I am afraid that we can only test the dependency on environments that we have.

For your information, here is the detailed env from conda env export on my server with
Linux 4.18.0-18-generic #19~18.04.1-Ubuntu x86_64 x86_64 x86_64 GNU/Linux and Nvidia Tesla p100:

name: pytorch-asvspoof2021
channels:
  - pytorch
  - conda-forge
  - defaults
dependencies:
  - _libgcc_mutex=0.1=conda_forge
  - _openmp_mutex=4.5=1_gnu
  - appdirs=1.4.4=py_0
  - audioread=2.1.9=py38h578d9bd_0
  - blas=1.0=mkl
  - brotlipy=0.7.0=py38h27cfd23_1003
  - bzip2=1.0.8=h7b6447c_0
  - ca-certificates=2021.5.25=h06a4308_1
  - certifi=2020.12.5=py38h06a4308_0
  - cffi=1.14.5=py38h261ae71_0
  - chardet=4.0.0=py38h06a4308_1003
  - cryptography=3.4.7=py38hd23ed53_0
  - cudatoolkit=9.2=0
  - cycler=0.10.0=py38_0
  - decorator=5.0.9=pyhd3eb1b0_0
  - ffmpeg=4.2.2=h20bf706_0
  - freetype=2.10.4=h5ab3b9f_0
  - future=0.18.2=py38_1
  - gettext=0.21.0=hf68c758_0
  - gmp=6.2.1=h2531618_2
  - gnutls=3.6.15=he1e5248_0
  - icu=58.2=he6710b0_3
  - idna=2.10=pyhd3eb1b0_0
  - intel-openmp=2021.2.0=h06a4308_610
  - joblib=1.0.1=pyhd3eb1b0_0
  - jpeg=9b=h024ee3a_2
  - kiwisolver=1.3.1=py38h2531618_0
  - lame=3.100=h7b6447c_0
  - lcms2=2.12=h3be6417_0
  - ld_impl_linux-64=2.33.1=h53a641e_7
  - libffi=3.3=he6710b0_2
  - libflac=1.3.3=h9c3ff4c_1
  - libgcc-ng=9.3.0=h2828fa1_19
  - libgfortran-ng=7.3.0=hdf63c60_0
  - libgomp=9.3.0=h2828fa1_19
  - libidn2=2.3.1=h27cfd23_0
  - libogg=1.3.4=h27cfd23_1
  - libopus=1.3.1=h7b6447c_0
  - libpng=1.6.37=hbc83047_0
  - librosa=0.8.0=pyh9f0ad1d_0
  - libsndfile=1.0.31=h9c3ff4c_1
  - libstdcxx-ng=9.3.0=h6de172a_19
  - libtasn1=4.16.0=h27cfd23_0
  - libtiff=4.2.0=h85742a9_0
  - libunistring=0.9.10=h27cfd23_0
  - libvorbis=1.3.7=h7b6447c_0
  - libvpx=1.7.0=h439df22_0
  - libwebp-base=1.2.0=h27cfd23_0
  - libxml2=2.9.10=hb55368b_3
  - llvmlite=0.31.0=py38hd408876_0
  - lz4-c=1.9.3=h2531618_0
  - matplotlib-base=3.3.4=py38h62a2d02_0
  - mir_eval=0.6=pyh9f0ad1d_0
  - mkl=2020.2=256
  - mkl-service=2.3.0=py38he904b0f_0
  - mkl_fft=1.3.0=py38h54f3939_0
  - mkl_random=1.1.1=py38h0573a6f_0
  - ncurses=6.2=he6710b0_1
  - nettle=3.7.2=hbbd107a_1
  - ninja=1.10.2=hff7bd54_1
  - numba=0.48.0=py38h0573a6f_0
  - numpy=1.18.1=py38h4f9e942_0
  - numpy-base=1.18.1=py38hde5b4d6_1
  - olefile=0.46=py_0
  - openh264=2.1.0=hd408876_0
  - openssl=1.1.1k=h27cfd23_0
  - packaging=20.9=pyhd3eb1b0_0
  - pillow=8.2.0=py38he98fc37_0
  - pip=21.1.1=py38h06a4308_0
  - pooch=1.3.0=pyhd3eb1b0_0
  - pycparser=2.20=py_2
  - pyopenssl=20.0.1=pyhd3eb1b0_1
  - pyparsing=2.4.7=pyhd3eb1b0_0
  - pysocks=1.7.1=py38h06a4308_0
  - pysoundfile=0.10.3.post1=pyhd3deb0d_0
  - python=3.8.10=hdb3f193_7
  - python-dateutil=2.8.1=pyhd3eb1b0_0
  - python_abi=3.8=1_cp38
  - pytorch=1.6.0=py3.8_cuda9.2.148_cudnn7.6.3_0
  - readline=8.1=h27cfd23_0
  - requests=2.25.1=pyhd3eb1b0_0
  - resampy=0.2.2=py_0
  - scikit-learn=0.24.2=py38ha9443f7_0
  - scipy=1.4.1=py38h0b6359f_0
  - setuptools=52.0.0=py38h06a4308_0
  - six=1.15.0=py38h06a4308_0
  - sqlite=3.35.4=hdfb4753_0
  - tbb=2021.2.0=hff7bd54_0
  - threadpoolctl=2.1.0=pyh5ca1d4c_0
  - tk=8.6.10=hbc83047_0
  - torchaudio=0.6.0=py38
  - torchvision=0.7.0=py38_cu92
  - tornado=6.1=py38h27cfd23_0
  - urllib3=1.26.4=pyhd3eb1b0_0
  - wheel=0.36.2=pyhd3eb1b0_0
  - x264=1!157.20191217=h7b6447c_0
  - xz=5.2.5=h7b6447c_0
  - zlib=1.2.11=h7b6447c_3
  - zstd=1.4.9=haebb681_0

[KyloRen1]

@TonyWangX

I have Windows 10 Enterprise with no GPU

And here is a part of my conflict log:

Package libflang conflicts for: pytorch::torchvision=0.7.0 -> numpy[version='>=1.11'] -> libflang[version='<6.0.0.a0|>=5.0.0'] librosa=0.8.0 -> numpy[version='>=1.15.0'] -> libflang[version='<6.0.0.a0|>=5.0.0'] numba=0.48.0 -> numpy[version='>=1.14.6,<2.0a0'] -> libflang[version='<6.0.0.a0|>=5.0.0'] mir_eval=0.6 -> numpy[version='>=1.7.0'] -> libflang[version='<6.0.0.a0|>=5.0.0'] pytorch::pytorch=1.6 -> numpy[version='>=1.11'] -> libflang[version='<6.0.0.a0|>=5.0.0'] conda-forge::pysoundfile -> numpy -> libflang[version='<6.0.0.a0|>=5.0.0'] scipy=1.4.1 -> numpy[version='>=1.14.6,<2.0a0'] -> libflang[version='<6.0.0.a0|>=5.0.0']

[TonyWangX]

Ahhh, sorry, I cannot help to debug since I don't have a Windows PC.

The issue seems to be due to libflang and numpy
Maybe related to this: conda/conda#9367?