diff --git a/tensorflow/contrib/lite/models/testdata/g3doc/README.md b/tensorflow/contrib/lite/models/testdata/g3doc/README.md
index 667a5883832914..1c47e00aae2a0e 100644
--- a/tensorflow/contrib/lite/models/testdata/g3doc/README.md
+++ b/tensorflow/contrib/lite/models/testdata/g3doc/README.md
@@ -53,7 +53,7 @@ with the corresponding parameters as shown in the figure.
 ### Automatic Speech Recognizer (ASR) Acoustic Model (AM)
 
 The acoustic model for automatic speech recognition is the neural network model
-for matching phonemes to the input autio features. It generates posterior
+for matching phonemes to the input audio features. It generates posterior
 probabilities of phonemes from speech frontend features (log-mel filterbanks).
 It has an input size of 320 (float), an output size of 42 (float), five LSTM
 layers and one fully connected layers with a Softmax activation function, with
@@ -68,7 +68,7 @@ for predicting the probability of a word given previous words in a sentence.
 It generates posterior probabilities of the next word based from a sequence of
 words. The words are encoded as indices in a fixed size dictionary.
 The model has two inputs both of size one (integer): the current word index and
-next word index, an output size of one (float): the log probability. It consits
+next word index, an output size of one (float): the log probability. It consists
 of three embedding layer, three LSTM layers, followed by a multiplication, a
 fully connected layers and an addition.
 The corresponding parameters as shown in the figure.
diff --git a/tensorflow/contrib/lite/toco/g3doc/cmdline_reference.md b/tensorflow/contrib/lite/toco/g3doc/cmdline_reference.md
index 4776741ab9273c..5e077952235fa1 100644
--- a/tensorflow/contrib/lite/toco/g3doc/cmdline_reference.md
+++ b/tensorflow/contrib/lite/toco/g3doc/cmdline_reference.md
@@ -229,7 +229,7 @@ additional information about the multiple input arrays:
     well-formed quantized representation of these graphs. Such graphs should be
     fixed, but as a temporary work-around, setting this
     reorder_across_fake_quant flag allows the converter to perform necessary
-    graph transformaitons on them, at the cost of no longer faithfully matching
+    graph transformations on them, at the cost of no longer faithfully matching
     inference and training arithmetic.
 
 ### Logging flags