nkululeko 0.80.1__tar.gz → 0.80.3__tar.gz

{nkululeko-0.80.1 → nkululeko-0.80.3}/CHANGELOG.md

@@ -1,6 +1,14 @@
 Changelog
 =========
 
+Version 0.80.3
+--------------
+* made sounddevice use optional as Portaudio library causes difficulties
+
+Version 0.80.2
+--------------
+* fixed bug that caused clash with GPU/CPU use
+
 Version 0.80.1
 --------------
 * added support for string value in import_features

{nkululeko-0.80.1/nkululeko.egg-info → nkululeko-0.80.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nkululeko
-Version: 0.80.1
+Version: 0.80.3
 Summary: Machine learning audio prediction experiments based on templates
 Home-page: https://github.com/felixbur/nkululeko
 Author: Felix Burkhardt
@@ -19,22 +19,18 @@ Requires-Dist: audinterface
 Requires-Dist: audiofile
 Requires-Dist: audiomentations
 Requires-Dist: audonnx
-Requires-Dist: cylimiter
 Requires-Dist: datasets
 Requires-Dist: imageio
-Requires-Dist: imbalanced-learn
 Requires-Dist: laion-clap
 Requires-Dist: matplotlib
 Requires-Dist: numpy
 Requires-Dist: opensmile
 Requires-Dist: pandas
 Requires-Dist: praat-parselmouth
-Requires-Dist: pylatex
 Requires-Dist: scikit_learn
 Requires-Dist: scipy
 Requires-Dist: seaborn
 Requires-Dist: sounddevice
-Requires-Dist: splitutils
 Requires-Dist: tensorflow
 Requires-Dist: tensorflow_hub
 Requires-Dist: torch
@@ -323,6 +319,14 @@ F. Burkhardt, Johannes Wagner, Hagen Wierstorf, Florian Eyben and Björn Schulle
 Changelog
 =========
 
+Version 0.80.3
+--------------
+* made sounddevice use optional as Portaudio library causes difficulties
+
+Version 0.80.2
+--------------
+* fixed bug that caused clash with GPU/CPU use
+
 Version 0.80.1
 --------------
 * added support for string value in import_features

{nkululeko-0.80.1 → nkululeko-0.80.3}/docs/source/conf.py

@@ -32,7 +32,9 @@ extensions = [
     'sphinx.ext.autodoc',
     'sphinx.ext.viewcode',
     # 'nbsphinx',           # to embedd ipynb files
-    'sphinx.ext.mathjax' # to enable latex
+    'sphinx.ext.mathjax', # to enable latex
+    'sphinx_rtd_theme',
+    'myst_parser',        # to enable markdown
 ]
 
 # Add any paths that contain templates here, relative to this directory.
@@ -54,10 +56,17 @@ html_theme = 'sphinx_rtd_theme'
 # Add any paths that contain custom static files (such as style sheets) here,
 # relative to this directory. They are copied after the builtin static files,
 # so a file named "default.css" will overwrite the builtin "default.css".
-html_static_path = ['_static']
+# html_static_path = ['_static']
 
 # suffix for source files
-source_suffix = '.rst'
+source_suffix = {
+    '.rst': 'restructuredtext',
+    '.txt': 'restructuredtext',
+    '.md': 'markdown',
+}
+
+# headed anchors until h3
+myst_heading_anchors = 3
 
 # The master toctree document.
 master_doc = 'index'

{nkululeko-0.80.1 → nkululeko-0.80.3}/nkululeko/constants.py

@@ -1,2 +1,2 @@
-VERSION="0.80.1"
+VERSION="0.80.3"
 SAMPLING_RATE = 16000

{nkululeko-0.80.1 → nkululeko-0.80.3}/nkululeko/demo_predictor.py

@@ -1,6 +1,5 @@
 import pandas as pd
 import numpy as np
-import sounddevice as sd
 import audiofile
 import nkululeko.glob_conf as glob_conf
 from nkululeko.utils.util import Util
@@ -84,6 +83,8 @@ class Demo_predictor:
             return dict_2
 
     def record_audio(self, seconds):
+        import sounddevice as sd
+
         print("recording ...")
         y = sd.rec(int(seconds * self.sr), samplerate=self.sr, channels=1)
         sd.wait()
@@ -91,6 +92,8 @@ class Demo_predictor:
         return y
 
     def play_audio(self, signal):
+        import sounddevice as sd
+
         print("playback ...")
         sd.play(signal.T, self.sr)
         status = sd.wait()

{nkululeko-0.80.1 → nkululeko-0.80.3}/nkululeko/feat_extract/feats_agender.py

@@ -18,6 +18,9 @@ class AudModelAgenderSet(Featureset):
 
     def __init__(self, name, data_df):
         super().__init__(name, data_df)
+        self.model_loaded = False
+
+    def _load_model(self):
         model_url = "https://zenodo.org/record/7761387/files/w2v2-L-robust-6-age-gender.25c844af-1.1.1.zip"
         model_root = self.util.config_val(
             "FEATS", "agender.model", "./audmodel_agender/"
@@ -25,12 +28,15 @@ class AudModelAgenderSet(Featureset):
         if not os.path.isdir(model_root):
             cache_root = audeer.mkdir("cache")
             model_root = audeer.mkdir(model_root)
-            archive_path = audeer.download_url(
-                model_url, cache_root, verbose=True
-            )
+            archive_path = audeer.download_url(model_url, cache_root, verbose=True)
             audeer.extract_archive(archive_path, model_root)
         device = self.util.config_val("MODEL", "device", "cpu")
         self.model = audonnx.load(model_root, device=device)
+        pytorch_total_params = sum(p.numel() for p in self.model.parameters())
+        self.util.debug(
+            f"initialized agender model with {pytorch_total_params} parameters in total"
+        )
+        self.model_loaded = True
 
     def extract(self):
         """Extract the features based on the initialized dataset or re-open them when found on disk."""
@@ -43,9 +49,10 @@ class AudModelAgenderSet(Featureset):
         no_reuse = eval(self.util.config_val("FEATS", "no_reuse", "False"))
         if no_reuse or extract or not os.path.isfile(storage):
             self.util.debug(
-                "extracting agender model embeddings, this might take a"
-                " while..."
+                "extracting agender model embeddings, this might take a" " while..."
             )
+            if not self.model_loaded:
+                self._load_model()
             hidden_states = audinterface.Feature(
                 self.model.labels("hidden_states"),
                 process_func=self.model,

{nkululeko-0.80.1 → nkululeko-0.80.3}/nkululeko/feat_extract/feats_agender_agender.py

@@ -18,6 +18,9 @@ class AgenderAgenderSet(Featureset):
 
     def __init__(self, name, data_df):
         super().__init__(name, data_df)
+        self.model_loaded = False
+
+    def _load_model(self):
         model_url = "https://zenodo.org/record/7761387/files/w2v2-L-robust-6-age-gender.25c844af-1.1.1.zip"
         model_root = self.util.config_val(
             "FEATS", "agender.model", "./audmodel_agender/"
@@ -33,6 +36,7 @@ class AgenderAgenderSet(Featureset):
         self.util.debug(
             f"initialized agender model with {pytorch_total_params} parameters in total"
         )
+        self.model_loaded = True
 
     def extract(self):
         """Extract the features based on the initialized dataset or re-open them when found on disk."""
@@ -48,6 +52,8 @@ class AgenderAgenderSet(Featureset):
             self.util.debug(
                 "extracting agender model age and gender, this might take a" " while..."
             )
+            if not self.model_loaded:
+                self._load_model()
             outputs = ["logits_age", "logits_gender"]
             logits = audinterface.Feature(
                 self.model.labels(outputs),

{nkululeko-0.80.1 → nkululeko-0.80.3}/nkululeko/feat_extract/feats_audmodel.py

@@ -20,6 +20,9 @@ class AudModelSet(Featureset):
 
     def __init__(self, name, data_df):
         super().__init__(name, data_df)
+        self.model_loaded = False
+
+    def _load_model(self):
         model_url = "https://zenodo.org/record/6221127/files/w2v2-L-robust-12.6bc4a7fd-1.1.0.zip"
         model_root = self.util.config_val("FEATS", "aud.model", "./audmodel/")
         if not os.path.isdir(model_root):
@@ -30,6 +33,7 @@ class AudModelSet(Featureset):
         cuda = "cuda" if torch.cuda.is_available() else "cpu"
         device = self.util.config_val("MODEL", "device", cuda)
         self.model = audonnx.load(model_root, device=device)
+        self.model_loaded = True
 
     def extract(self):
         """Extract the features based on the initialized dataset or re-open them when found on disk."""
@@ -44,6 +48,8 @@ class AudModelSet(Featureset):
             self.util.debug(
                 "extracting audmodel embeddings, this might take a while..."
             )
+            if not self.model_loaded:
+                self._load_model()
             hidden_states = audinterface.Feature(
                 self.model.labels("hidden_states"),
                 process_func=self.model,

{nkululeko-0.80.1 → nkululeko-0.80.3}/nkululeko/feat_extract/feats_audmodel_dim.py

@@ -18,6 +18,9 @@ class AudModelDimSet(Featureset):
 
     def __init__(self, name, data_df):
         super().__init__(name, data_df)
+        self.model_loaded = False
+
+    def _load_model(self):
         model_url = "https://zenodo.org/record/6221127/files/w2v2-L-robust-12.6bc4a7fd-1.1.0.zip"
         model_root = self.util.config_val("FEATS", "aud.model", "./audmodel/")
         if not os.path.isdir(model_root):
@@ -25,8 +28,10 @@ class AudModelDimSet(Featureset):
             model_root = audeer.mkdir(model_root)
             archive_path = audeer.download_url(model_url, cache_root, verbose=True)
             audeer.extract_archive(archive_path, model_root)
-        device = self.util.config_val("MODEL", "device", "cpu")
+        cuda = "cuda" if torch.cuda.is_available() else "cpu"
+        device = self.util.config_val("MODEL", "device", cuda)
         self.model = audonnx.load(model_root, device=device)
+        self.model_loaded = True
 
     def extract(self):
         """Extract the features based on the initialized dataset or re-open them when found on disk."""
@@ -41,6 +46,8 @@ class AudModelDimSet(Featureset):
             self.util.debug(
                 "extracting audmodel dimensions, this might take a while..."
             )
+            if not self.model_loaded:
+                self._load_model()
             logits = audinterface.Feature(
                 self.model.labels("logits"),
                 process_func=self.model,

{nkululeko-0.80.1 → nkululeko-0.80.3}/nkululeko/models/model_mlp.py

@@ -107,8 +107,8 @@ class MLP_model(Model):
                 logits[start_index:end_index, :] = model(features.to(device))
                 targets[start_index:end_index] = labels
                 loss = self.criterion(
-                    logits[start_index:end_index, :],
-                    labels.to(self.device, dtype=torch.int64),
+                    logits[start_index:end_index, :].to(device),
+                    labels.to(device, dtype=torch.int64),
                 )
                 losses.append(loss.item())
 

{nkululeko-0.80.1 → nkululeko-0.80.3}/nkululeko/models/model_mlp_regression.py

@@ -185,7 +185,12 @@ class MLP_Reg_model(Model):
                     end_index = len(loader.dataset)
                 logits[start_index:end_index] = model(features.to(device)).reshape(-1)
                 targets[start_index:end_index] = labels
-                loss = self.criterion(logits[start_index:end_index], labels.to(device))
+                loss = self.criterion(
+                    logits[start_index:end_index].to(
+                        device,
+                    ),
+                    labels.to(device),
+                )
                 losses.append(loss.item())
         self.loss_eval = (np.asarray(losses)).mean()
 

{nkululeko-0.80.1 → nkululeko-0.80.3/nkululeko.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nkululeko
-Version: 0.80.1
+Version: 0.80.3
 Summary: Machine learning audio prediction experiments based on templates
 Home-page: https://github.com/felixbur/nkululeko
 Author: Felix Burkhardt
@@ -19,22 +19,18 @@ Requires-Dist: audinterface
 Requires-Dist: audiofile
 Requires-Dist: audiomentations
 Requires-Dist: audonnx
-Requires-Dist: cylimiter
 Requires-Dist: datasets
 Requires-Dist: imageio
-Requires-Dist: imbalanced-learn
 Requires-Dist: laion-clap
 Requires-Dist: matplotlib
 Requires-Dist: numpy
 Requires-Dist: opensmile
 Requires-Dist: pandas
 Requires-Dist: praat-parselmouth
-Requires-Dist: pylatex
 Requires-Dist: scikit_learn
 Requires-Dist: scipy
 Requires-Dist: seaborn
 Requires-Dist: sounddevice
-Requires-Dist: splitutils
 Requires-Dist: tensorflow
 Requires-Dist: tensorflow_hub
 Requires-Dist: torch
@@ -323,6 +319,14 @@ F. Burkhardt, Johannes Wagner, Hagen Wierstorf, Florian Eyben and Björn Schulle
 Changelog
 =========
 
+Version 0.80.3
+--------------
+* made sounddevice use optional as Portaudio library causes difficulties
+
+Version 0.80.2
+--------------
+* fixed bug that caused clash with GPU/CPU use
+
 Version 0.80.1
 --------------
 * added support for string value in import_features

{nkululeko-0.80.1 → nkululeko-0.80.3}/nkululeko.egg-info/requires.txt

@@ -4,22 +4,18 @@ audinterface
 audiofile
 audiomentations
 audonnx
-cylimiter
 datasets
 imageio
-imbalanced-learn
 laion-clap
 matplotlib
 numpy
 opensmile
 pandas
 praat-parselmouth
-pylatex
 scikit_learn
 scipy
 seaborn
 sounddevice
-splitutils
 tensorflow
 tensorflow_hub
 torch

{nkululeko-0.80.1 → nkululeko-0.80.3}/setup.cfg

@@ -23,22 +23,18 @@ install_requires =
 	audiofile
 	audiomentations
 	audonnx
-	cylimiter
 	datasets
 	imageio
-	imbalanced-learn
 	laion-clap
 	matplotlib
 	numpy
 	opensmile
 	pandas
 	praat-parselmouth
-	pylatex
 	scikit_learn
 	scipy
 	seaborn
 	sounddevice
-	splitutils
 	tensorflow
 	tensorflow_hub
 	torch