Merge branch 'dev_al' of https://git-lium.univ-lemans.fr/Larcher/sidekit into dev_al

features_server.py

@@ -488,6 +488,7 @@ class FeaturesServer(object):
             feat, label = self.post_processing(feat, label, global_mean, global_std)
         else:
             feat, label = self.post_processing(feat, label)
+
         return feat, label
 
     def get_features_per_speaker(self, show, idmap, channel=0, input_feature_filename=None, label=None):

nnet/__init__.py

@@ -48,14 +48,7 @@ from .preprocessor import RawPreprocessor
 from .preprocessor import MfccFrontEnd 
 from .preprocessor import MelSpecFrontEnd
 
-has_pyroom = True
-try:
-   import pyroomacoustics
-except ImportError:
-   has_pyroom = False
 
-if has_pyroom:
-    from .augmentation import AddReverb
 
 
 __author__ = "Anthony Larcher and Sylvain Meignier"

nnet/xvector.py

@@ -1480,17 +1480,17 @@ def xtrain(dataset_description,
                                                           validation_non_indices,
                                                           training_opts["mixed_precision"])
 
-            test_eer = None
-            if training_opts["compute_test_eer"] and local_rank < 1:
-                test_eer = test_metrics(model, device, model_opts, dataset_opts, training_opts)
+            #test_eer = None
+            #if training_opts["compute_test_eer"] and local_rank < 1:
+            #    test_eer = test_metrics(model, device, model_opts, dataset_opts, training_opts)
 
             monitor.update(test_eer=test_eer,
                            val_eer=val_eer,
                            val_loss=val_loss,
                            val_acc=val_acc)
 
-            if local_rank < 1:
-                monitor.display()
+            #if local_rank < 1:
+            #    monitor.display()
 
             # Save the current model and if needed update the best one
             # TODO ajouter une option qui garde les modèles à certaines époques (par exemple avant le changement de LR
@@ -1675,6 +1675,7 @@ def extract_embeddings(idmap_name,
                        model_filename,
                        data_root_name,
                        device,
+                       batch_size=1,
                        file_extension="wav",
                        transform_pipeline={},
                        sliding_window=False,
@@ -1700,6 +1701,10 @@ def extract_embeddings(idmap_name,
     :param mixed_precision:
     :return:
     """
+
+    if sliding_window:
+        batch_size = 1
+
     # Load the model
     if isinstance(model_filename, str):
         checkpoint = torch.load(model_filename, map_location=device)
@@ -1729,7 +1734,7 @@ def extract_embeddings(idmap_name,
                        )
 
     dataloader = DataLoader(dataset,
-                            batch_size=1,
+                            batch_size=batch_size,
                             shuffle=False,
                             drop_last=False,
                             pin_memory=True,