Script to convert to jasper format

custom_formats.sh

@@ -0,0 +1,10 @@
+# This is the path where all output is stored
+export PYTHONPATH=$PYTHONPATH:$(pwd)/src
+out_path=data
+
+echo "##############################################################"
+echo "# Jasperize"
+echo "##############################################################"
+wave_path=$out_path/full_waverized
+jasper_path=$out_path/full_jasperized
+python scripts/jasperize.py $wave_path $jasper_path

data/.gitignore

@@ -2,3 +2,4 @@ download/
 full/
 full_normalized/
 full_waverized/
+full_jasperized/

requirements.txt

@@ -1,4 +1,4 @@
 click==7.0
 tqdm==4.39.0
-git+git://github.com/ynop/audiomate.git@92f5725#egg=audiomate
+git+git://github.com/ynop/audiomate.git@30488ce#egg=audiomate
 git+git://github.com/ynop/spoteno.git@7700d53#egg=spoteno

scripts/jasperize.py

@@ -0,0 +1,34 @@
+import os
+import click
+
+import audiomate
+from audiomate.corpus import io
+
+
+@click.command()
+@click.argument('in_folder', type=click.Path(exists=True))
+@click.argument('out_folder', type=click.Path())
+@click.option('--base-folder', default=None, type=click.Path())
+def run(in_folder, out_folder, base_folder):
+    if not os.path.exists(out_folder):
+        if base_folder is None:
+            base_folder = os.path.dirname(out_folder)
+
+        w = io.NvidiaJasperWriter(
+            num_workers=8,
+            no_check=True,
+            data_base_path=base_folder
+        )
+        target_audio_path = os.path.join(out_folder, 'audio')
+        os.makedirs(target_audio_path)
+
+        print('Load source corpus')
+        ds = audiomate.Corpus.load(in_folder)
+        print('Save jasper corpus')
+        w.save(ds, out_folder)
+    else:
+        print('Already jasperized')
+
+
+if __name__ == '__main__':
+    run()