diff --git a/DeepSpeech.py b/DeepSpeech.py
index ea9503da..f470b520 100755
--- a/DeepSpeech.py
+++ b/DeepSpeech.py
@@ -30,7 +30,7 @@ from util.feeding import DataSet, ModelFeeder
 from util.preprocess import preprocess
 from util.gpu import get_available_gpus
 from util.shared_lib import check_cupti
-from util.text import sparse_tensor_value_to_texts, wer, levenshtein, Alphabet, ndarray_to_text
+from util.text import sparse_tensor_value_to_texts, wer, levenshtein, Alphabet
 from xdg import BaseDirectory as xdg
 import numpy as np
 
diff --git a/util/text.py b/util/text.py
index 96bd1805..5caf5fae 100644
--- a/util/text.py
+++ b/util/text.py
@@ -10,6 +10,7 @@ from functools import reduce
 
 class Alphabet(object):
     def __init__(self, config_file):
+        self._config_file = config_file
         self._label_to_str = []
         self._str_to_label = {}
         self._size = 0
@@ -29,9 +30,18 @@ class Alphabet(object):
     def label_from_string(self, string):
         return self._str_to_label[string]
 
+    def decode(self, labels):
+        res = ''
+        for label in labels:
+            res += self.string_from_label(label)
+        return res
+
     def size(self):
         return self._size
 
+    def config_file(self):
+        return self._config_file
+
 def text_to_char_array(original, alphabet):
     r"""
     Given a Python string ``original``, remove unsupported characters, map characters
@@ -76,12 +86,6 @@ def sparse_tuple_to_texts(tuple, alphabet):
     # List of strings
     return results
 
-def ndarray_to_text(value, alphabet):
-    results = ''
-    for i in range(len(value)):
-        results += alphabet.string_from_label(value[i])
-    return results
-
 def wer(original, result):
     r"""
     The WER is defined as the editing/Levenshtein distance on word level