beamforming.py uses removed torch.complex32

ben-freist commented 2 years ago

Before reporting a bug:

First, please search previous issues and the FAQ and be sure this hasn't been answered elsewhere.

🐛 Bug

torch temporarily removed torch32 for some fixes (https://github.com/pytorch/pytorch/issues/72721), it'll be brought back in a later version.

To Reproduce

Steps to reproduce the behavior (code sample and stack trace):

checkout the asteroid repo
pip install -e .

python -m pytest tests/dsp/beamforming_test.py

============================================================================================ short test summary info =============================================================================================
ERROR tests/dsp/beamforming_test.py - AttributeError: module 'torch' has no attribute 'complex32'
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Interrupted: 1 error during collection !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
================================================================================================ 1 error in 2.00s ================================================================================================

Expected behavior

The test should pass.

Environment

With python 3.6 an older torch version is used but both python3.7.12 and python3.9.10 give. me this problem.

Package versions

My commit is 25b03c7d158ae30469c5e1b4bffd14d8cdf70e21 on master.

john7rho commented 2 years ago

hello, i was wondering if there were any work-arounds for this? i believe the speech separation also does not work on versions past python3.6, but i may be mistaken.

mpariente commented 2 years ago

Yes, sorry, I haven't looked into this issue. I'll have a look in the coming weeks.

mpariente commented 2 years ago

Closed by #609

asteroid-team / asteroid