akadmin/openwakeword
1
0
Fork 0
mirror of https://github.com/dscripka/openWakeWord.git synced 2026-08-27 18:17:20 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Passing flake8 and mypy tests locally [skip ci]

Browse source
This commit is contained in:
dscripka 2023-10-01 21:40:09 -04:00
parent fd36a564cd
commit a07006136f
2 changed files with 11 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

6
openwakeword/data.py
View file

@ -451,7 +451,7 @@ def mix_clips_batch(
# Apply volume augmentation
if volume_augmentation:
volume_levels = np.random.uniform(0.02, 1.0, mixed_clips_batch.shape[0])
mixed_clips_batch = (volume_levels/mixed_clips_batch.max(axis=1)[0])[..., None]*mixed_clips_batch
mixed_clips_batch = (volume_levels/mixed_clips_batch.max(dim=1)[0])[..., None]*mixed_clips_batch
else:
# Normalize clips only if max value is outside of [-1, 1]
abs_max, _ = torch.max(
@ -463,7 +463,7 @@ def mix_clips_batch(
mixed_clips_batch = (mixed_clips_batch.numpy()*32767).astype(np.int16)
# Remove any clips that are silent (happens rarely when mixing/reverberating)
error_index = np.where(mixed_clips_batch.max(axis=1) != 0)[0]
error_index = torch.from_numpy(np.where(mixed_clips_batch.max(dim=1) != 0)[0])
mixed_clips_batch = mixed_clips_batch[error_index]
labels_batch = labels_batch[error_index]
sequence_labels_batch = sequence_labels_batch[error_index]
@ -686,7 +686,7 @@ def augment_clips(
# Do second pass augmentations
device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')
augmented_batch = augment2(samples=torch.vstack(augmented_clips).unsqueeze(axis=1).to(device), sample_rate=sr).squeeze(axis=1)
augmented_batch = augment2(samples=torch.vstack(augmented_clips).unsqueeze(dim=1).to(device), sample_rate=sr).squeeze(axis=1)
# Do reverberation
if augmentation_probabilities["RIR"] >= np.random.random() and RIR_paths != []:

14
openwakeword/train.py
View file

@ -410,6 +410,7 @@ class Model(nn.Module):
if step_ndx == max_steps-1:
break
# Separate function to convert onnx models to tflite format
def convert_onnx_to_tflite(onnx_model_path, output_path):
"""Converts an ONNX version of an openwakeword model to the Tensorflow tflite format."""
@ -697,14 +698,15 @@ if __name__ == '__main__':
batch_size=len(X_val_fp_labels)
)
X_val = np.vstack((
np.load(os.path.join(feature_save_dir, "positive_features_test.npy")),
np.load(os.path.join(feature_save_dir, "negative_features_test.npy"))
))
labels = np.hstack((np.ones(X_val.shape[0]//2), np.zeros(X_val.shape[0]//2))).astype(np.float32)
X_val_pos = np.load(os.path.join(feature_save_dir, "positive_features_test.npy"))
X_val_neg = np.load(os.path.join(feature_save_dir, "negative_features_test.npy"))
labels = np.hstack((np.ones(X_val_pos.shape[0]), np.zeros(X_val_neg.shape[0]))).astype(np.float32)
X_val = torch.utils.data.DataLoader(
torch.utils.data.TensorDataset(torch.from_numpy(X_val), torch.from_numpy(labels)),
torch.utils.data.TensorDataset(
torch.from_numpy(np.vstack((X_val_pos, X_val_neg))),
torch.from_numpy(labels)
),
batch_size=len(labels)
)