nkululeko 0.90.0__py3-none-any.whl → 0.90.2__py3-none-any.whl

nkululeko/aug_train.py

@@ -1,15 +1,17 @@
 # aug_train.py
 # train with augmentations
+import argparse
 import ast
+import configparser
 import os.path
+
 import numpy as np
-import configparser
-import argparse
+
 import nkululeko.experiment as exp
-from nkululeko.utils.util import Util
-from nkululeko.constants import VERSION
 import nkululeko.glob_conf as glob_conf
 from nkululeko.augment import doit as augment
+from nkululeko.constants import VERSION
+from nkululeko.utils.util import Util
 
 
 def doit(config_file):

nkululeko/augment.py

@@ -2,13 +2,15 @@
 # augment the training sets
 
 import argparse
-import pandas as pd
-import os
 import ast
-from nkululeko.experiment import Experiment
 import configparser
-from nkululeko.utils.util import Util
+import os
+
+import pandas as pd
+
 from nkululeko.constants import VERSION
+from nkululeko.experiment import Experiment
+from nkululeko.utils.util import Util
 
 
 def doit(config_file):

nkululeko/augmenting/augmenter.py

@@ -1,12 +1,12 @@
 # augmenter.py
 import os
-import numpy as np
+
+import audeer
+import audiofile
 import pandas as pd
 from audiomentations import *
 from tqdm import tqdm
-import audeer
-import audiofile
-from audformat.utils import map_file_path
+
 from nkululeko.utils.util import Util
 
 

nkululeko/augmenting/randomsplicer.py

@@ -13,15 +13,15 @@ F. Burkhardt, Anna Derington, Matthias Kahlau, Klaus Scherer, Florian Eyben and
 
 """
 
-import pandas as pd
-from tqdm import tqdm
 import os
-import numpy as np
-import audiofile as af
-from audformat.utils import map_file_path
+
 import audeer
-from nkululeko.utils.util import Util
+import audiofile as af
+import pandas as pd
+from tqdm import tqdm
+
 import nkululeko.augmenting.randomsplicing as rsp
+from nkululeko.utils.util import Util
 
 
 class Randomsplicer:

nkululeko/augmenting/randomsplicing.py

@@ -10,9 +10,9 @@ Evaluated in:
 F. Burkhardt, Anna Derington, Matthias Kahlau, Klaus Scherer, Florian Eyben and Björn Schuller: Masking Speech Contents by Random Splicing: is Emotional Expression Preserved?, Proc. ICASSP, 2023
 
 """
-import numpy as np
+
 import librosa
-import audiofile as af
+import numpy as np
 
 
 def random_splicing(
@@ -54,7 +54,6 @@ def split_wav_naive(wav, top_db=12):
 
 
 def remix_random_reverse(wav, indices, p_reverse=0):
-    import random
 
     wav_remix = []
 

nkululeko/augmenting/resampler.py

@@ -2,12 +2,14 @@
 resample a data frame
 
 """
+
 import os
 import shutil
 
 import audformat
 import pandas as pd
 import torchaudio
+
 from nkululeko.utils.util import Util
 
 
@@ -18,8 +20,11 @@ class Resampler:
         self.util = Util("resampler", has_config=not_testing)
         self.util.warn(f"all files might be resampled to {self.SAMPLING_RATE}")
         self.not_testing = not_testing
-        self.replace = eval(self.util.config_val(
-            "RESAMPLE", "replace", "False")) if not not_testing else replace
+        self.replace = (
+            eval(self.util.config_val("RESAMPLE", "replace", "False"))
+            if not not_testing
+            else replace
+        )
 
     def resample(self):
         files = self.df.index.get_level_values(0).values
@@ -45,8 +50,7 @@ class Resampler:
                 continue
             if org_sr != self.SAMPLING_RATE:
                 self.util.debug(f"resampling {f} (sr = {org_sr})")
-                resampler = torchaudio.transforms.Resample(
-                    org_sr, self.SAMPLING_RATE)
+                resampler = torchaudio.transforms.Resample(org_sr, self.SAMPLING_RATE)
                 signal = resampler(signal)
                 if replace:
                     torchaudio.save(
@@ -63,8 +67,7 @@ class Resampler:
             self.df = self.df.set_index(
                 self.df.index.set_levels(new_files, level="file")
             )
-            target_file = self.util.config_val(
-                "RESAMPLE", "target", "resampled.csv")
+            target_file = self.util.config_val("RESAMPLE", "target", "resampled.csv")
             # remove encoded labels
             target = self.util.config_val("DATA", "target", "emotion")
             if "class_label" in self.df.columns:

nkululeko/autopredict/ap_age.py

@@ -2,10 +2,12 @@
 A predictor for age.
 Currently based on audEERING's agender model.
 """
-from nkululeko.utils.util import Util
-from nkululeko.feature_extractor import FeatureExtractor
+
 import ast
+
 import nkululeko.glob_conf as glob_conf
+from nkululeko.feature_extractor import FeatureExtractor
+from nkululeko.utils.util import Util
 
 
 class AgePredictor:

nkululeko/autopredict/ap_arousal.py

@@ -2,10 +2,12 @@
 A predictor for emotional arousal.
 Currently based on audEERING's emotional dimension model.
 """
-from nkululeko.utils.util import Util
-from nkululeko.feature_extractor import FeatureExtractor
+
 import ast
+
 import nkululeko.glob_conf as glob_conf
+from nkululeko.feature_extractor import FeatureExtractor
+from nkululeko.utils.util import Util
 
 
 class ArousalPredictor:

nkululeko/autopredict/ap_dominance.py

@@ -3,10 +3,11 @@ A predictor for emotional dominance.
 Currently based on audEERING's emotional dimension model.
 """
 
-from nkululeko.utils.util import Util
-from nkululeko.feature_extractor import FeatureExtractor
 import ast
+
 import nkululeko.glob_conf as glob_conf
+from nkululeko.feature_extractor import FeatureExtractor
+from nkululeko.utils.util import Util
 
 
 class DominancePredictor:

nkululeko/autopredict/ap_gender.py

@@ -2,10 +2,12 @@
 A predictor for biological sex.
 Currently based on audEERING's agender model.
 """
-from nkululeko.utils.util import Util
-from nkululeko.feature_extractor import FeatureExtractor
+
 import ast
+
 import nkululeko.glob_conf as glob_conf
+from nkululeko.feature_extractor import FeatureExtractor
+from nkululeko.utils.util import Util
 
 
 class GenderPredictor:

nkululeko/autopredict/ap_mos.py

@@ -1,11 +1,14 @@
 """"
 A predictor for MOS - mean opinion score.
 """
-from nkululeko.utils.util import Util
+
 import ast
+
+import numpy as np
+
 import nkululeko.glob_conf as glob_conf
 from nkululeko.feature_extractor import FeatureExtractor
-import numpy as np
+from nkululeko.utils.util import Util
 
 
 class MOSPredictor:

nkululeko/autopredict/ap_pesq.py

@@ -1,11 +1,14 @@
 """"
 A predictor for PESQ - Perceptual Evaluation of Speech Quality.
 """
-from nkululeko.utils.util import Util
+
 import ast
+
+import numpy as np
+
 import nkululeko.glob_conf as glob_conf
 from nkululeko.feature_extractor import FeatureExtractor
-import numpy as np
+from nkululeko.utils.util import Util
 
 
 class PESQPredictor:

nkululeko/autopredict/ap_sdr.py

@@ -2,11 +2,14 @@
 A predictor for SDR - Signal to Distortion Ratio.
 as estimated by Scale-Invariant Signal-to-Distortion Ratio (SI-SDR) 
 """
-from nkululeko.utils.util import Util
+
 import ast
+
+import numpy as np
+
 import nkululeko.glob_conf as glob_conf
 from nkululeko.feature_extractor import FeatureExtractor
-import numpy as np
+from nkululeko.utils.util import Util
 
 
 class SDRPredictor:

nkululeko/autopredict/ap_snr.py

@@ -1,11 +1,14 @@
 """"
 A predictor for SNR - signal-to-noise ratio.
 """
-from nkululeko.utils.util import Util
+
 import ast
+
+import numpy as np
+
 import nkululeko.glob_conf as glob_conf
 from nkululeko.feature_extractor import FeatureExtractor
-import numpy as np
+from nkululeko.utils.util import Util
 
 
 class SNRPredictor:

nkululeko/autopredict/ap_stoi.py

@@ -1,11 +1,14 @@
 """"
 A predictor for STOI - Short-Time Objective Intelligibility (STOI)
 """
-from nkululeko.utils.util import Util
+
 import ast
+
+import numpy as np
+
 import nkululeko.glob_conf as glob_conf
 from nkululeko.feature_extractor import FeatureExtractor
-import numpy as np
+from nkululeko.utils.util import Util
 
 
 class STOIPredictor:

nkululeko/autopredict/ap_valence.py

@@ -2,10 +2,12 @@
 A predictor for emotional valence.
 Currently based on audEERING's emotional dimension model.
 """
-from nkululeko.utils.util import Util
-from nkululeko.feature_extractor import FeatureExtractor
+
 import ast
+
 import nkululeko.glob_conf as glob_conf
+from nkululeko.feature_extractor import FeatureExtractor
+from nkululeko.utils.util import Util
 
 
 class ValencePredictor:

nkululeko/autopredict/estimate_snr.py

@@ -43,9 +43,7 @@ class SNREstimator:
     def frame_audio(self, signal):
         num_frames = 1 + (len(signal) - self.frame_length) // self.hop_length
         frames = [
-            signal[
-                i * self.hop_length: (i * self.hop_length) + self.frame_length
-            ]
+            signal[i * self.hop_length : (i * self.hop_length) + self.frame_length]
             for i in range(num_frames)
         ]
         return frames
@@ -64,11 +62,8 @@ class SNREstimator:
             for frame in frames
         ]
 
-        energy_threshold_low = np.percentile(
-            log_energies, 25)  # First quartile
-        energy_threshold_high = np.percentile(
-            log_energies, 75
-        )  # Third quartile
+        energy_threshold_low = np.percentile(log_energies, 25)  # First quartile
+        energy_threshold_high = np.percentile(log_energies, 75)  # Third quartile
 
         low_energy_frames = [
             log_energy
@@ -94,9 +89,7 @@ class SNREstimator:
             energy_threshold_high,
         )
 
-    def plot_energy(
-        self, log_energies, energy_threshold_low, energy_threshold_high
-    ):
+    def plot_energy(self, log_energies, energy_threshold_low, energy_threshold_high):
         plt.figure(figsize=(10, 6))
         plt.plot(log_energies, label="Log Energy")
         plt.axhline(
@@ -153,9 +146,12 @@ def main():
 
     signal, sr = audiofile.read(args.input)
     snr_estimator = SNREstimator(signal, sr, args.window_size, args.hop_size)
-    estimated_snr, log_energies, energy_threshold_low, energy_threshold_high = (
-        snr_estimator.estimate_snr()
-    )
+    (
+        estimated_snr,
+        log_energies,
+        energy_threshold_low,
+        energy_threshold_high,
+    ) = snr_estimator.estimate_snr()
 
     print("Estimated SNR:", estimated_snr)
 

nkululeko/cacheddataset.py

@@ -1,5 +1,5 @@
-import torch
 import numpy as np
+import torch
 
 
 class CachedDataset(torch.utils.data.Dataset):

nkululeko/constants.py

@@ -1,2 +1,2 @@
-VERSION="0.90.0"
+VERSION = "0.90.2"
 SAMPLING_RATE = 16000

nkululeko/data/dataset.py

@@ -3,13 +3,11 @@ import ast
 import os
 import os.path
 from random import sample
-import numpy as np
-import pandas as pd
 
 import audformat
-from audformat.utils import duration
+import numpy as np
+import pandas as pd
 
-import nkululeko.filter_data as filter
 import nkululeko.glob_conf as glob_conf
 from nkululeko.filter_data import DataFilter
 from nkululeko.plots import Plots
@@ -274,20 +272,20 @@ class Dataset:
                 # try to get the target values
                 df_local[self.target] = source_df[self.col_label]
                 is_labeled = True
-            except (KeyError, ValueError, audformat.errors.BadKeyError) as e:
+            except (KeyError, ValueError, audformat.errors.BadKeyError):
                 pass
             try:
                 # try to get the speaker values
                 df_local["speaker"] = source_df["speaker"]
                 got_speaker = True
-            except (KeyError, ValueError, audformat.errors.BadKeyError) as e:
+            except (KeyError, ValueError, audformat.errors.BadKeyError):
                 pass
             try:
                 # try to get the gender values
                 if "gender" in source_df:
                     df_local["gender"] = source_df["gender"]
                     got_gender = True
-            except (KeyError, ValueError, audformat.errors.BadKeyError) as e:
+            except (KeyError, ValueError, audformat.errors.BadKeyError):
                 pass
             try:
                 # try to get the age values
@@ -311,13 +309,13 @@ class Dataset:
                 # also it might be possible that the age is part of the speaker description
                 df_local["age"] = db[table]["speaker"].get(map="age").astype(int)
                 got_age = True
-            except (ValueError, audformat.errors.BadKeyError) as e:
+            except (ValueError, audformat.errors.BadKeyError):
                 pass
             try:
                 # same for the target, e.g. "age"
                 df_local[self.target] = db[table]["speaker"].get(map=self.target)
                 is_labeled = True
-            except (ValueError, audformat.core.errors.BadKeyError) as e:
+            except (ValueError, audformat.core.errors.BadKeyError):
                 pass
             # copy other column
             for column in source_df.columns:
@@ -463,8 +461,7 @@ class Dataset:
 
     def balanced_split(self):
         """One way to split train and eval sets: Generate split dataframes for some balancing criterion"""
-        from splitutils import binning
-        from splitutils import optimize_traintest_split
+        from splitutils import binning, optimize_traintest_split
 
         seed = 42
         k = 30
@@ -473,7 +470,7 @@ class Dataset:
         # split target
         targets = df[self.target].to_numpy()
         #
-        bins = self.util.config_val("DATA", f"bin", False)
+        bins = self.util.config_val("DATA", "bin", False)
         if bins:
             nbins = len(ast.literal_eval(bins))
             targets = binning(targets, nbins=nbins)
@@ -481,7 +478,7 @@ class Dataset:
         speakers = df["speaker"].to_numpy()
 
         # on which variables (targets, groupings) to stratify
-        stratif_vars = self.util.config_val("DATA", f"balance", False)
+        stratif_vars = self.util.config_val("DATA", "balance", False)
         stratif_vars_array = {}
         if not stratif_vars:
             self.util.error("balanced split needs stratif_vars to stratify the splits")
@@ -500,7 +497,7 @@ class Dataset:
         # weights for all stratify_on variables and
         # and for test proportion match. Give target
         # variable EMOTION more weight than groupings.
-        size_diff = int(self.util.config_val("DATA", f"size_diff_weight", "1"))
+        size_diff = int(self.util.config_val("DATA", "size_diff_weight", "1"))
         weights = {
             "size_diff": size_diff,
         }

nkululeko/data/dataset_csv.py

@@ -2,10 +2,12 @@
 import ast
 import os
 import os.path
-import pandas as pd
+
 import audformat.utils
-from nkululeko.data.dataset import Dataset
+import pandas as pd
+
 import nkululeko.glob_conf as glob_conf
+from nkululeko.data.dataset import Dataset
 from nkululeko.reporting.report_item import ReportItem
 
 
@@ -81,7 +83,7 @@ class Dataset_CSV(Dataset):
         self.start_fresh = eval(self.util.config_val("DATA", "no_reuse", "False"))
         is_index = False
         try:
-            if self.is_labeled and not "class_label" in self.df.columns:
+            if self.is_labeled and "class_label" not in self.df.columns:
                 self.df["class_label"] = self.df[self.target]
         except AttributeError:
             is_index = True

nkululeko/demo-ft.py

@@ -0,0 +1,29 @@
+import argparse
+import configparser
+import os
+
+from sklearn import pipeline
+from transformers import pipelines
+
+from nkululeko.utils.util import get_exp_dir
+
+parser = argparse.ArgumentParser()
+parser.add_argument("--file", help="A file that should be processed (16kHz mono wav)")
+# read config from ini file
+parser.add_argument("--config", default="exp.ini", help="The base configuration")
+
+
+args = parser.parse_args()
+file = args.file
+config = configparser.ConfigParser()
+
+# get exp dir from config [EXP][root][name] + models + run_0 + torch
+config.read(args.config)
+exp_dir = get_exp_dir("model_path")
+
+# exp_dir = get_exp_dir("model_path")
+model_path = os.path.join(exp_dir, "model")
+pipe = pipelines("audio-classification", model=model_path)
+
+
+print(pipeline(file))

nkululeko/demo_feats.py

@@ -1,13 +1,14 @@
 # demo_feats.py
 # Test some features extracted
 
-from nkululeko.experiment import Experiment
-import configparser
-from nkululeko.utils.util import Util
-from nkululeko.constants import VERSION
 import argparse
+import configparser
 import os
 
+from nkululeko.constants import VERSION
+from nkululeko.experiment import Experiment
+from nkululeko.utils.util import Util
+
 
 def main(src_dir):
     parser = argparse.ArgumentParser(description="Call the nkululeko framework.")

nkululeko/demo_predictor.py

@@ -1,11 +1,10 @@
 # demo_predictor.py
 import os
 
-import numpy as np
-import pandas as pd
-
 import audformat
 import audiofile
+import numpy as np
+import pandas as pd
 
 import nkululeko.glob_conf as glob_conf
 from nkululeko.utils.util import Util
@@ -46,7 +45,7 @@ class Demo_predictor:
                         )
                     else:
                         file_list = in_df.index.values
-                except (ValueError, AttributeError) as error:
+                except (ValueError, AttributeError):
                     with open(self.file) as f:
                         first = True
                         for index, line in enumerate(f):

nkululeko/ensemble.py

@@ -15,28 +15,20 @@ Raises:
     ValueError: If an unknown ensemble method is provided.
     AssertionError: If the number of config files is less than 2 for majority voting.
 """
+
 #!/usr/bin/env python
 # -*- coding: utf-8 -*-
 
 
-from typing import List
 import configparser
 import time
 from argparse import ArgumentParser
 from pathlib import Path
+from typing import List
 
 import numpy as np
 import pandas as pd
-import matplotlib.pyplot as plt
-
-from sklearn.metrics import(
-    RocCurveDisplay,
-    balanced_accuracy_score, 
-    classification_report,
-    auc,
-    roc_auc_score,
-    roc_curve
-)
+from sklearn.metrics import balanced_accuracy_score, classification_report
 
 from nkululeko.constants import VERSION
 from nkululeko.experiment import Experiment
@@ -169,17 +161,19 @@ def performance_weighted_ensemble(ensemble_preds_ls, labels, weights):
 
     # asserts weiths in decimal 0-1
     assert all(0 <= w <= 1 for w in weights), "Weights must be between 0 and 1"
-    
+
     # assert lenght of weights matches number of models
-    assert len(weights) == len(ensemble_preds_ls), "Number of weights must match number of models"
-    
+    assert len(weights) == len(
+        ensemble_preds_ls
+    ), "Number of weights must match number of models"
+
     # Normalize weights
     total_weight = sum(weights)
     weights = [weight / total_weight for weight in weights]
-    
+
     for idx in ensemble_preds_ls[0].index:
         class_probabilities = {label: 0 for label in labels}
-        
+
         for df, weight in zip(ensemble_preds_ls, weights):
             row = df.loc[idx]
             for label in labels:
@@ -192,10 +186,12 @@ def performance_weighted_ensemble(ensemble_preds_ls, labels, weights):
     return final_predictions, final_confidences
 
 
-
-
 def ensemble_predictions(
-    config_files: List[str], method: str, threshold: float, weights: List[float], no_labels: bool
+    config_files: List[str],
+    method: str,
+    threshold: float,
+    weights: List[float],
+    no_labels: bool,
 ) -> pd.DataFrame:
     """
     Ensemble predictions from multiple experiments.
@@ -261,17 +257,20 @@ def ensemble_predictions(
             ensemble_preds_ls, labels, threshold
         )
     elif method == "uncertainty_weighted":
-        ensemble_preds["predicted"], ensemble_preds["uncertainty"] = (
-            uncertainty_weighted_ensemble(ensemble_preds_ls, labels)
-        )
+        (
+            ensemble_preds["predicted"],
+            ensemble_preds["uncertainty"],
+        ) = uncertainty_weighted_ensemble(ensemble_preds_ls, labels)
     elif method == "confidence_weighted":
-        ensemble_preds["predicted"], ensemble_preds["confidence"] = (
-            confidence_weighted_ensemble(ensemble_preds_ls, labels)
-        )
+        (
+            ensemble_preds["predicted"],
+            ensemble_preds["confidence"],
+        ) = confidence_weighted_ensemble(ensemble_preds_ls, labels)
     elif method == "performance_weighted":
-        ensemble_preds["predicted"], ensemble_preds["confidence"] = (
-            performance_weighted_ensemble(ensemble_preds_ls, labels, weights)
-        )
+        (
+            ensemble_preds["predicted"],
+            ensemble_preds["confidence"],
+        ) = performance_weighted_ensemble(ensemble_preds_ls, labels, weights)
     else:
         raise ValueError(f"Unknown ensemble method: {method}")