Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Adding test for T.SlidingWindowCmn #1482

Merged
merged 5 commits into from
May 3, 2021
Merged

Adding test for T.SlidingWindowCmn #1482

Changes from 1 commit
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
d10930e
Adding test for T.SlidingWindowCmn
pavithranrao Apr 30, 2021
6ec8994
Removing one negative case
pavithranrao Apr 30, 2021
96e4606
Removing duplicate test case
pavithranrao Apr 30, 2021
a279859
Adding failing test cases for
pavithranrao May 1, 2021
c5256e5
Using spectrogram instead of waveform for T.SlidingWindowCmn
pavithranrao May 3, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 0 additions & 1 deletion test/torchaudio_unittest/transforms/autograd_test_impl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -157,7 +157,6 @@ def test_vol(self, gain, gain_type):
@parameterized.expand([
({'cmn_window': 600, 'min_cmn_window': 100, 'center': False, 'norm_vars': False}, ),
({'cmn_window': 600, 'min_cmn_window': 100, 'center': True, 'norm_vars': False}, ),
({'cmn_window': 600, 'min_cmn_window': 100, 'center': False, 'norm_vars': True}, ),
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@mthrok this set of params causes the test to fail with RuntimeError: Jacobian mismatch for output 0 with respect to input 0. Any idea why?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This means that when norm_vars=True, some operation is not differential to the 2nd degree.

It's somewhere here but it is not immediately clear to me.

audio/torchaudio/functional/functional.py

Lines 1073 to 1100 in 0c263a9

if norm_vars:
cur_sumsq += torch.cumsum(input_part ** 2, 1)[:, -1, :]
else:
if window_start > last_window_start:
frame_to_remove = specgram[:, last_window_start, :]
cur_sum -= frame_to_remove
if norm_vars:
cur_sumsq -= (frame_to_remove ** 2)
if window_end > last_window_end:
frame_to_add = specgram[:, last_window_end, :]
cur_sum += frame_to_add
if norm_vars:
cur_sumsq += (frame_to_add ** 2)
window_frames = window_end - window_start
last_window_start = window_start
last_window_end = window_end
cmn_specgram[:, t, :] = specgram[:, t, :] - cur_sum / window_frames
if norm_vars:
if window_frames == 1:
cmn_specgram[:, t, :] = torch.zeros(
num_channels, num_feats, dtype=dtype, device=device)
else:
variance = cur_sumsq
variance = variance / window_frames
variance -= ((cur_sum ** 2) / (window_frames ** 2))
variance = torch.pow(variance, -0.5)
cmn_specgram[:, t, :] *= variance

What we want to do is

  1. Identify which part is causing this.
  2. Change the code if that is possible without performance degradation.

However, they are beyond the scope of this PR, so here, we can set nondet_tol and add docstring saying it's not 2nd-order differentiable when norm_vars=True, like in Spectrogram.

audio/test/torchaudio_unittest/transforms/autograd_test_impl.py

Lines 71 to 77 in 0c263a9

# replication_pad1d_backward_cuda is not deteministic and
# gives very small (~2.7756e-17) difference.
#
# See https://github.com/pytorch/pytorch/issues/54093
transform = T.Spectrogram(**kwargs)
waveform = get_whitenoise(sample_rate=8000, duration=0.05, n_channels=2)
self.assert_grad(transform, [waveform], nondet_tol=1e-10)

({'cmn_window': 600, 'min_cmn_window': 100, 'center': False, 'norm_vars': False}, ),
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this set of params is a duplicate of the first one -- could you change or remove it?

Copy link
Contributor Author

@pavithranrao pavithranrao Apr 30, 2021 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for catching that, will change remove it.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Removed the duplicate test case. Thanks!

({'cmn_window': 500, 'min_cmn_window': 50, 'center': False, 'norm_vars': False}, ),
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

cmn_window =600 and min_cmn_window=100 look too big for the input with 8000 * 0.05 == 400 (then FFT applied) can you make them somewhat smaller than the number of frames in time axis of the input tensor?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

thanks @mthrok will try to implement your suggestions

])
Expand Down