Remove last old manifest usage.

Signed-off-by: Stanislav Beliaev <stasbelyaev96@gmail.com>
NVIDIA · Jan 22, 2020 · 00a9d59 · 00a9d59
1 parent b3a8cd4
commit 00a9d59
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 17 deletions.
diff --git a/collections/nemo_asr/nemo_asr/parts/manifest.py b/collections/nemo_asr/nemo_asr/parts/manifest.py
@@ -9,15 +9,16 @@
 
 class ManifestBase():
     def __init__(self,
-                 manifest_paths,
-                 labels,
-                 max_duration=None,
-                 min_duration=None,
-                 sort_by_duration=False,
-                 max_utts=0,
-                 blank_index=-1,
-                 unk_index=-1,
-                 normalize=True):
+         manifest_paths,
+         labels,
+         max_duration=None,
+         min_duration=None,
+         sort_by_duration=False,
+         max_utts=0,
+         blank_index=-1,
+         unk_index=-1,
+         normalize=True,
+    ):
         self.min_duration = min_duration
         self.max_duration = max_duration
         self.sort_by_duration = sort_by_duration
@@ -61,7 +62,8 @@ def __init__(self,
 
             # tokenize transcript text
             item["tokens"] = self.tokenize_transcript(
-                    text, self.labels_map, self.unk_index, self.blank_index)
+                text, self.labels_map, self.unk_index, self.blank_index
+            )
 
             # support files using audio_filename
             if 'audio_filename' in item and 'audio_filepath' not in item:
@@ -155,11 +157,7 @@ def normalize_text(text, labels):
         # Punctuation to remove
         punctuation = string.punctuation
         # Define punctuation that will be handled by text cleaner
-        punctuation_to_replace = {
-            "+": "plus",
-            "&": "and",
-            "%": "percent"
-        }
+        punctuation_to_replace = {"+": "plus", "&": "and", "%": "percent"}
         for char in punctuation_to_replace:
             punctuation = punctuation.replace(char, "")
         # We might also want to consider:

diff --git a/collections/nemo_asr/nemo_asr/parts/perturb.py b/collections/nemo_asr/nemo_asr/parts/perturb.py
@@ -3,6 +3,8 @@
 import random
 
 import librosa
+from nemo_asr.parts import char_parsers
+from nemo_asr.parts import manifests
 from scipy import signal
 
 from .segment import AudioSegment
@@ -47,7 +49,9 @@ def perturb(self, data):
 
 class ImpulsePerturbation(Perturbation):
     def __init__(self, manifest_path=None, rng=None):
-        self._manifest = ManifestEN(manifest_path)
+        self._manifest = manifests.ASRAudioText(
+            manifest_path, parser=char_parsers.make_parser([])
+        )
         self._rng = random.Random() if rng is None else rng
 
     def perturb(self, data):
@@ -91,7 +95,9 @@ def __init__(
         max_gain_db=300.0,
         rng=None,
     ):
-        self._manifest = ManifestEN(manifest_path)
+        self._manifest = manifests.ASRAudioText(
+            manifest_path, parser=char_parsers.make_parser([])
+        )
         self._rng = random.Random() if rng is None else rng
         self._min_snr_db = min_snr_db
         self._max_snr_db = max_snr_db