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

Jump to bottom

Ignore the NoUtterancesError when calculating pesq for a batch #2753

Open
wants to merge 9 commits into
base: master
Choose a base branch
from
Open

Ignore the NoUtterancesError when calculating pesq for a batch #2753

Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
d2aa54f
filter nan
veera-puthiran-14082 Sep 17, 2024
f08cca8
use np
veera-puthiran-14082 Sep 17, 2024
10f1e06
fix tests
veera-puthiran-14082 Sep 17, 2024
852299a
Merge branch 'master' into fix_pesq_nan
veera-puthiran-14082 Sep 17, 2024
434c105
fix dtype from object
veera-puthiran-14082 Sep 17, 2024
07b8a69
Merge branch 'master' into fix_pesq_nan
Borda Sep 17, 2024
30551b7
Merge branch 'master' into fix_pesq_nan
veera-puthiran-14082 Sep 18, 2024
68b3e49
Merge branch 'master' into fix_pesq_nan
veera-puthiran-14082 Sep 24, 2024
a079851
Merge branch 'master' into fix_pesq_nan
veera-puthiran-14082 Sep 26, 2024
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
6 changes: 4 additions & 2 deletions src/torchmetrics/functional/audio/pesq.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -83,6 +83,8 @@ def perceptual_evaluation_speech_quality(
)
import pesq as pesq_backend

_filter_error_msg = np.vectorize(lambda x: np.issubdtype(type(x), np.number))

if fs not in (8000, 16000):
raise ValueError(f"Expected argument `fs` to either be 8000 or 16000 but got {fs}")
if mode not in ("wb", "nb"):
Expand All @@ -103,8 +105,8 @@ def perceptual_evaluation_speech_quality(
pesq_val_np = np.empty(shape=(preds_np.shape[0]))
for b in range(preds_np.shape[0]):
pesq_val_np[b] = pesq_backend.pesq(fs, target_np[b, :], preds_np[b, :], mode)
pesq_val = torch.from_numpy(pesq_val_np)
pesq_val = pesq_val.reshape(preds.shape[:-1])
pesq_val = torch.from_numpy(pesq_val_np[_filter_error_msg(pesq_val_np)].astype(np.float32))
pesq_val = pesq_val.reshape(len(pesq_val))

if keep_same_device:
return pesq_val.to(preds.device)
Expand Down