get_integrated_lufs now concatenates files <0.5 before computing LUFS

To avoid getting constant -70.0 for files shorter than 400 ms

scaper/audio.py

@@ -6,7 +6,11 @@
 '''
 
 import subprocess
+import sox
+import numpy as np
+import tempfile
 from .scaper_exceptions import ScaperError
+from .util import _close_temp_files
 
 
 def r128stats(filepath):
@@ -50,8 +54,53 @@ def r128stats(filepath):
     return stats_dict
 
 
-def get_integrated_lufs(filepath):
-    '''Returns the integrated lufs for an audiofile'''
+def get_integrated_lufs(filepath, min_duration=0.5):
+    """
+    Returns the integrated LUFS for an audiofile.
+
+    For files shorter than 400 ms ffmpeg returns a constant integrated LUFS
+    value of -70.0. To avoid this, files shorter than min_duration (by default
+    500 ms) are self-concatenated until min_duration is reached and the
+    LUFS value is computed for the concatenated file.
+
+    Parameters
+    ----------
+    filepath : str
+        Path to audio file for computing LUFS
+    min_duration : float
+        Minimum required duration for computing LUFS value. Files shorter than
+        this are self-concatenated until their duration reaches this value
+        for the purpose of computing the integrated LUFS. Caution: if you set
+        min_duration < 0.4, a constant LUFS value of -70.0 will be returned for
+        all files shorter than 400 ms.
+
+    Returns
+    -------
+
+    """
+    try:
+        duration = sox.file_info.duration(filepath)
+    except Exception as e:
+        raise ScaperError(
+            'Unable to obtain LUFS for {:s}, error message:\n{:s}'.format(
+                filepath, e.__str__()))
+
+    if duration < min_duration:
+        # compute how many concatenations we require
+        n_tiles = int(np.ceil(min_duration / duration))
+
+        # Concatenate audio to itself, save to temp file and get LUFS
+        tmpfiles = []
+        with _close_temp_files(tmpfiles):
+            concat_file = tempfile.NamedTemporaryFile(suffix='.wav',
+                                                      delete=True)
+            tmpfiles.append(concat_file)
+
+            cbn = sox.Combiner()
+            cbn.build([filepath] * n_tiles, concat_file.name, 'concatenate')
+
+            loudness_stats = r128stats(concat_file.name)
+    else:
+        loudness_stats = r128stats(filepath)
 
-    loudness_stats = r128stats(filepath)
     return loudness_stats['I']