Remove manifest's line number check from librispeech.py and update README.md.

README.md

@@ -22,6 +22,10 @@ cat manifest.libri.train-* > manifest.libri.train-all
 cd ..
 ```
 
+After running librispeech.py, we have several "manifest" json files named with a prefix `manifest.libri.`. A manifest file summarizes a speech data set, with each line containing the meta data (i.e. audio filepath, transcription text, audio duration) of each audio file within the data set, in json format.
+
+By `cat manifest.libri.train-* > manifest.libri.train-all`, we simply merge the three seperate sample sets of LibriSpeech (train-clean-100, train-clean-360, train-other-500) into one training set. This is a simple way for merging different data sets.
+
 More help for arguments:
 
 ```

data/librispeech.py

@@ -1,10 +1,9 @@
 """
-    Download, unpack and create manifest file for the Librespeech dataset.
+    Download, unpack and create manifest json files for the Librespeech dataset.
 
-    A manifest file is a dataset summarization, with each line a json format
+    A manifest is a json file summarizing filelist in a data set, with each line
-    string containing meta data for one audio clip, including its filepath,
+    containing the meta data (i.e. audio filepath, transcription text, audio
-    transcription string, and duration. It serves as a unified interface for
+    duration) of each audio file in the data set.
-    different data sets.
 """
 
 import paddle.v2 as paddle
@@ -36,14 +35,6 @@ MD5_TRAIN_CLEAN_100 = "2a93770f6d5c6c964bc36631d331a522"
 MD5_TRAIN_CLEAN_360 = "c0e676e450a7ff2f54aeade5171606fa"
 MD5_TRAIN_OTHER_500 = "d1a0fd59409feb2c614ce4d30c387708"
 
-NUM_LINES_TEST_CLEAN = 2620
-NUM_LINES_TEST_OTHER = 2939
-NUM_LINES_DEV_CLEAN = 2703
-NUM_LINES_DEV_OTHER = 2864
-NUM_LINES_TRAIN_CLEAN_100 = 28539
-NUM_LINES_TRAIN_CLEAN_360 = 104014
-NUM_LINES_TRAIN_OTHER_500 = 148688
-
 parser = argparse.ArgumentParser(
     description='Downloads and prepare LibriSpeech dataset.')
 parser.add_argument(
@@ -95,12 +86,9 @@ def unpack(filepath, target_dir):
 
 def create_manifest(data_dir, manifest_path):
     """
-    Create a manifest file summarizing the dataset (list of filepath and meta
+    Create a manifest json file summarizing the data set, with each line
-    data).
+    containing the meta data (i.e. audio filepath, transcription text, audio
-
+    duration) of each audio file within the data set.
-    Each line of the manifest contains one audio clip filepath, its
-    transcription text string, and its duration. Manifest file servers as a
-    unified interfance to organize data sets.
     """
     print("Creating manifest %s ..." % manifest_path)
     json_lines = []
@@ -128,28 +116,20 @@ def create_manifest(data_dir, manifest_path):
             out_file.write(line + '\n')
 
 
-def verify_file_line_number(filepath, num_lines):
+def prepare_dataset(url, md5sum, target_dir, manifest_path):
-    with open(filepath, 'r') as file:
-        return len(file.readlines()) == num_lines
-
-
-def prepare_dataset(url, md5sum, target_dir, manifest_path, num_lines):
     """
     Download, unpack and create summmary manifest file.
     """
+    if not os.path.exists(os.path.join(target_dir, "LibriSpeech")):
         # download
         filepath = download(url, md5sum, target_dir)
         # unpack
-    if not os.path.exists(os.path.join(target_dir, "LibriSpeech")):
         unpack(filepath, target_dir)
     else:
-        print("Unpacked data exists, skip unpacking.")
+        print("Skip downloading and unpacking. Data already exists in %s." %
-    # create manifest and verify line number
+              target_dir)
+    # create manifest json file
     create_manifest(target_dir, manifest_path)
-    if not verify_file_line_number(manifest_path, num_lines):
-        raise RuntimeError("Manifest line number check failed. "
-                           "Please remove directory and try running the script "
-                           "again.")
 
 
 def main():
@@ -157,45 +137,38 @@ def main():
         url=URL_TEST_CLEAN,
         md5sum=MD5_TEST_CLEAN,
         target_dir=os.path.join(args.target_dir, "test-clean"),
-        manifest_path=args.manifest_prefix + ".test-clean",
+        manifest_path=args.manifest_prefix + ".test-clean")
-        num_lines=NUM_LINES_TEST_CLEAN)
     prepare_dataset(
         url=URL_DEV_CLEAN,
         md5sum=MD5_DEV_CLEAN,
         target_dir=os.path.join(args.target_dir, "dev-clean"),
-        manifest_path=args.manifest_prefix + ".dev-clean",
+        manifest_path=args.manifest_prefix + ".dev-clean")
-        num_lines=NUM_LINES_DEV_CLEAN)
     prepare_dataset(
         url=URL_TRAIN_CLEAN_100,
         md5sum=MD5_TRAIN_CLEAN_100,
         target_dir=os.path.join(args.target_dir, "train-clean-100"),
-        manifest_path=args.manifest_prefix + ".train-clean-100",
+        manifest_path=args.manifest_prefix + ".train-clean-100")
-        num_lines=NUM_LINES_TRAIN_CLEAN_100)
     if args.full_download:
         prepare_dataset(
             url=URL_TEST_OTHER,
             md5sum=MD5_TEST_OTHER,
             target_dir=os.path.join(args.target_dir, "test-other"),
-            manifest_path=args.manifest_prefix + ".test-other",
+            manifest_path=args.manifest_prefix + ".test-other")
-            num_lines=NUM_LINES_TEST_OTHER)
         prepare_dataset(
             url=URL_DEV_OTHER,
             md5sum=MD5_DEV_OTHER,
             target_dir=os.path.join(args.target_dir, "dev-other"),
-            manifest_path=args.manifest_prefix + ".dev-other",
+            manifest_path=args.manifest_prefix + ".dev-other")
-            num_lines=NUM_LINES_DEV_OTHER)
         prepare_dataset(
             url=URL_TRAIN_CLEAN_360,
             md5sum=MD5_TRAIN_CLEAN_360,
             target_dir=os.path.join(args.target_dir, "train-clean-360"),
-            manifest_path=args.manifest_prefix + ".train-clean-360",
+            manifest_path=args.manifest_prefix + ".train-clean-360")
-            num_lines=NUM_LINES_TRAIN_CLEAN_360)
         prepare_dataset(
             url=URL_TRAIN_OTHER_500,
             md5sum=MD5_TRAIN_OTHER_500,
             target_dir=os.path.join(args.target_dir, "train-other-500"),
-            manifest_path=args.manifest_prefix + ".train-other-500",
+            manifest_path=args.manifest_prefix + ".train-other-500")
-            num_lines=NUM_LINES_TRAIN_OTHER_500)
 
 
 if __name__ == '__main__':