Merge branch 'develop' of https://github.com/PaddlePaddle/DeepSpeech into readme_translation

README.md

@@ -539,4 +539,4 @@ We compare the training time with 1, 2, 4, 8, 16 Tesla K40m GPUs (with a subset
 
 ## Questions and Help
 
-You are welcome to submit questions and bug reports in [Github Issues](https://github.com/PaddlePaddle/models/issues). You are also welcome to contribute to this project.
+You are welcome to submit questions and bug reports in [Github Issues](https://github.com/PaddlePaddle/DeepSpeech/issues). You are also welcome to contribute to this project.

data_utils/data.py

@@ -11,7 +11,6 @@ import multiprocessing
 import numpy as np
 import paddle.v2 as paddle
 from threading import local
-import atexit
 from data_utils.utility import read_manifest
 from data_utils.utility import xmap_readers_mp
 from data_utils.augmentor.augmentation import AugmentationPipeline
@@ -194,15 +193,18 @@ class DataGenerator(object):
                     raise ValueError("Unknown shuffle method %s." %
                                      shuffle_method)
             # prepare batches
-            instance_reader = self._instance_reader_creator(manifest)
+            instance_reader, cleanup = self._instance_reader_creator(manifest)
             batch = []
-            for instance in instance_reader():
+            try:
-                batch.append(instance)
+                for instance in instance_reader():
-                if len(batch) == batch_size:
+                    batch.append(instance)
+                    if len(batch) == batch_size:
+                        yield self._padding_batch(batch, padding_to, flatten)
+                        batch = []
+                if len(batch) >= min_batch_size:
                     yield self._padding_batch(batch, padding_to, flatten)
-                    batch = []
+            finally:
-            if len(batch) >= min_batch_size:
+                cleanup()
-                yield self._padding_batch(batch, padding_to, flatten)
             self._epoch += 1
 
         return batch_reader
@@ -280,10 +282,7 @@ class DataGenerator(object):
             lambda instance: self.process_utterance(instance["audio_filepath"], instance["text"]),
             reader, self._num_threads, 4096)
 
-        # register callback to main process
+        return reader, cleanup_callback
-        atexit.register(cleanup_callback)
-
-        return reader
 
     def _padding_batch(self, batch, padding_to=-1, flatten=False):
         """