nkululeko 0.90.4__tar.gz → 0.91.0__tar.gz

{nkululeko-0.90.4 → nkululeko-0.91.0}/CHANGELOG.md

@@ -1,6 +1,11 @@
 Changelog
 =========
 
+Version 0.91.0
+--------------
+* fixed duration display in segmentation
+* added possibility to use original segmentations (without max. duration)
+
 Version 0.90.4
 --------------
 * added plot format for multidb

{nkululeko-0.90.4/nkululeko.egg-info → nkululeko-0.91.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nkululeko
-Version: 0.90.4
+Version: 0.91.0
 Summary: Machine learning audio prediction experiments based on templates
 Home-page: https://github.com/felixbur/nkululeko
 Author: Felix Burkhardt
@@ -356,6 +356,11 @@ F. Burkhardt, Johannes Wagner, Hagen Wierstorf, Florian Eyben and Björn Schulle
 Changelog
 =========
 
+Version 0.91.0
+--------------
+* fixed duration display in segmentation
+* added possibility to use original segmentations (without max. duration)
+
 Version 0.90.4
 --------------
 * added plot format for multidb

{nkululeko-0.90.4 → nkululeko-0.91.0}/nkululeko/constants.py

@@ -1,2 +1,2 @@
-VERSION="0.90.4"
+VERSION="0.91.0"
 SAMPLING_RATE = 16000

{nkululeko-0.90.4 → nkululeko-0.91.0}/nkululeko/segment.py

@@ -7,9 +7,9 @@ import os
 
 import pandas as pd
 
-import nkululeko.glob_conf as glob_conf
 from nkululeko.constants import VERSION
 from nkululeko.experiment import Experiment
+import nkululeko.glob_conf as glob_conf
 from nkululeko.reporting.report_item import ReportItem
 from nkululeko.utils.util import Util
 
@@ -78,6 +78,7 @@ def main():
 
     if "duration" not in df.columns:
         df["duration"] = df.index.to_series().map(lambda x: calc_dur(x))
+    df_seg["duration"] = df_seg.index.to_series().map(lambda x: calc_dur(x))
     num_before = df.shape[0]
     num_after = df_seg.shape[0]
     # plot distributions
@@ -115,36 +116,5 @@ def main():
     print("DONE")
 
 
-def get_segmentation(file):
-    #    print(f'segmenting {file[0]}')
-    print(".", end="")
-    wav = read_audio(file[0], sampling_rate=SAMPLING_RATE)
-    speech_timestamps = get_speech_timestamps(
-        wav, vad_model, sampling_rate=SAMPLING_RATE
-    )
-    files, starts, ends = [], [], []
-    for entry in speech_timestamps:
-        start = float(entry["start"] / 1000.0)
-        end = float(entry["end"] / 1000.0)
-        files.append(file[0])
-        starts.append(start)
-        ends.append(end)
-    seg_index = segmented_index(files, starts, ends)
-    return seg_index
-
-
-def segment_dataframe(df):
-    dfs = []
-    for file, values in df.iterrows():
-        index = get_segmentation(file)
-        dfs.append(
-            pd.DataFrame(
-                values.to_dict(),
-                index,
-            )
-        )
-    return audformat.utils.concat(dfs)
-
-
 if __name__ == "__main__":
     main()  # use this if you want to state the config file path on command line

{nkululeko-0.90.4 → nkululeko-0.91.0}/nkululeko/segmenting/seg_silero.py

@@ -32,8 +32,30 @@ class Silero_segmenter:
         self.no_testing = not_testing
         self.util = Util(has_config=not_testing)
 
-    def get_segmentation(self, file):
-        #    print(f'segmenting {file[0]}')
+    def get_segmentation_simple(self, file):
+        (
+            get_speech_timestamps,
+            save_audio,
+            read_audio,
+            VADIterator,
+            collect_chunks,
+        ) = vad_utils
+        SAMPLING_RATE = 16000
+        wav = read_audio(file[0], sampling_rate=SAMPLING_RATE)
+        speech_timestamps = get_speech_timestamps(
+            wav, vad_model, sampling_rate=SAMPLING_RATE
+        )
+        files, starts, ends = [], [], []
+        for entry in speech_timestamps:
+            start = float(entry["start"] / SAMPLING_RATE)
+            end = float(entry["end"] / SAMPLING_RATE)
+            files.append(file[0])
+            starts.append(start)
+            ends.append(end)
+        seg_index = segmented_index(files, starts, ends)
+        return seg_index
+
+    def get_segmentation(self, file, min_length, max_length):
         (
             get_speech_timestamps,
             save_audio,
@@ -42,12 +64,6 @@ class Silero_segmenter:
             collect_chunks,
         ) = vad_utils
         SAMPLING_RATE = 16000
-        if self.no_testing:
-            min_length = float(self.util.config_val("SEGMENT", "min_length", 2))
-            max_length = float(self.util.config_val("SEGMENT", "max_length", 10))
-        else:
-            min_length = 2
-            max_length = 10
         wav = read_audio(file[0], sampling_rate=SAMPLING_RATE)
         speech_timestamps = get_speech_timestamps(
             wav, vad_model, sampling_rate=SAMPLING_RATE
@@ -76,8 +92,18 @@ class Silero_segmenter:
 
     def segment_dataframe(self, df):
         dfs = []
+        max_length = eval(self.util.config_val("SEGMENT", "max_length", "False"))
+        if max_length:
+            if self.no_testing:
+                min_length = float(self.util.config_val("SEGMENT", "min_length", 2))
+            else:
+                min_length = 2
+            self.util.debug(f"segmenting with max length: {max_length+min_length}")
         for file, values in tqdm(df.iterrows()):
-            index = self.get_segmentation(file)
+            if max_length:
+                index = self.get_segmentation(file, min_length, max_length)
+            else:
+                index = self.get_segmentation_simple(file)
             dfs.append(
                 pd.DataFrame(
                     values.to_dict(),

{nkululeko-0.90.4 → nkululeko-0.91.0/nkululeko.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nkululeko
-Version: 0.90.4
+Version: 0.91.0
 Summary: Machine learning audio prediction experiments based on templates
 Home-page: https://github.com/felixbur/nkululeko
 Author: Felix Burkhardt
@@ -356,6 +356,11 @@ F. Burkhardt, Johannes Wagner, Hagen Wierstorf, Florian Eyben and Björn Schulle
 Changelog
 =========
 
+Version 0.91.0
+--------------
+* fixed duration display in segmentation
+* added possibility to use original segmentations (without max. duration)
+
 Version 0.90.4
 --------------
 * added plot format for multidb