Merge pull request fgnt#3 from TCord/main

Fix usage of scenario_key for LibriSpeech database

mms_msg/databases/single_speaker/librispeech/database.py

@@ -1,3 +1,5 @@
+from typing import Iterable
+
 from lazy_dataset.database import JsonDatabase
 from mms_msg import keys
 from mms_msg.simulation.utils import load_audio
@@ -16,13 +18,15 @@ def load_example(self, example):
 
     def add_scenario(self, example):
         example['scenario'] = '_'.join([example[key] for key in sorted(self.scenario_key)])
+        return example
 
     def get_dataset(self, name=None):
         dataset = super().get_dataset(name)
         if self.scenario_key is not None:
-            dataset.map(self.add_scenario)
+            dataset = dataset.map(self.add_scenario)
         return dataset
 
+
 class LibriSpeech8kHz(LibriSpeech):
-    def __init__(self, json_path=database_jsons / 'librispeech_8k.json'):
-        super().__init__(json_path)
+    def __init__(self, json_path=database_jsons / 'librispeech_8k.json', scenario_key=None):
+        super().__init__(json_path, scenario_key=scenario_key)

mms_msg/databases/utils.py

@@ -33,31 +33,9 @@ def check_audio_files_exist(
         speedup=None,
 ):
     """
+    Verifies all files denoted in a dabatase_dict as paths actually exist.
     No structure for the database_dict is assumed. It will just search for all
     string values ending with a certain file type (e.g. wav).
-
-    >>> check_audio_files_exist({2: [1, '/net/db/timit/pcm/train/dr1/fcjf0/sa1.wav', 'abc.wav']})
-    Traceback (most recent call last):
-      ...
-    AssertionError: ('abc.wav', (2, '2'))
-    >>> check_audio_files_exist(1)
-    Traceback (most recent call last):
-      ...
-    AssertionError: Expect at least one wav file. It is likely that the database folder is empty and the greps did not work. to_check: {}
-    >>> check_audio_files_exist('abc.wav')
-    Traceback (most recent call last):
-      ...
-    AssertionError: ('abc.wav', ())
-    >>> check_audio_files_exist('/net/db/timit/pcm/train/dr1/fcjf0/sa1.wav')
-    >>> check_audio_files_exist(1, speedup='thread')
-    Traceback (most recent call last):
-      ...
-    AssertionError: Expect at least one wav file. It is likely that the database folder is empty and the greps did not work. to_check: {}
-    >>> check_audio_files_exist('abc.wav', speedup='thread')
-    Traceback (most recent call last):
-      ...
-    AssertionError: ('abc.wav', ())
-    >>> check_audio_files_exist('/net/db/timit/pcm/train/dr1/fcjf0/sa1.wav', speedup='thread')
     """
 
     def path_exists(path):