diff --git a/README.md b/README.md
index 1b6c65e..930ef91 100644
--- a/README.md
+++ b/README.md
@@ -1,23 +1,28 @@
 # Subsync
+
 **Synchronize your subtitles using machine learning**
 
 Subsync analyses and processes the sound from your media files and uses machine learning to detect speech. Speech detection is used to shift existing subtitles for a perfect match in audio and text!
 
 ## Features
- - [x] Machine learning model for voice activity detection (*not recognition*)
- - [x] Shift subtitle as a whole for best match
- - [x] Sync every sentence in the subtitle individually
- - [ ] Sync using existing matched subtitle in a different laguage
+
+- [x] Machine learning model for voice activity detection (*not recognition*)
+- [x] Shift subtitle as a whole for best match
+- [x] Sync every sentence in the subtitle individually
+- [ ] Sync using existing matched subtitle in a different laguage
 
 ## Dependencies
-* ffmpeg (https://www.ffmpeg.org/download.html)
+
+* ffmpeg (<https://www.ffmpeg.org/download.html>)
 
 ## Installation
+
 ```bash
 pip install subsync
 ```
 
 ## Help
+
 ```
 usage: subsync [-h] [--version] [--graph] [-d SECONDS] [-m SECONDS] [-s]
                    [--logfile PATH]
@@ -46,4 +51,5 @@ optional arguments:
 ```
 
 ## Special thanks
-[[1] Automatic Subtitle Synchronization through Machine Learning](https://machinelearnings.co/automatic-subtitle-synchronization-e188a9275617) 
+
+[[1] Automatic Subtitle Synchronization through Machine Learning](https://machinelearnings.co/automatic-subtitle-synchronization-e188a9275617)
diff --git a/requirements.txt b/requirements.txt
index 444bfae..3544402 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -3,18 +3,18 @@ astor==0.6.2
 audioread==2.1.5
 bleach==1.5.0
 cffi==1.11.5
-chardet==3.0.4
+chardet>=3.0.4
 cycler==0.10.0
 decorator==4.2.1
 gast==0.2.0
 graphviz==0.8.2
 grpcio==1.10.0
-h5py==2.8.0rc1
+h5py>2.8.0
 html5lib==0.9999999
 joblib==0.11
 Keras==2.1.5
 kiwisolver==1.0.1
-librosa==0.6.0
+librosa>=0.6.0
 llvmlite==0.22.0
 Markdown==2.6.11
 matplotlib==2.2.2
@@ -29,12 +29,12 @@ python-dateutil==2.7.0
 pytz==2018.3
 PyYAML==3.12
 resampy==0.2.0
-scikit-learn==0.19.1
+scikit-learn>=0.19.1
 scipy==1.0.0
 six==1.11.0
 sklearn==0.0
 tensorboard==1.6.0
-tensorflow==1.5.0
-tensorflow-tensorboard==1.5.1
+tensorflow>1.5.0
+tensorflow-tensorboard>=1.5.1
 termcolor==1.1.0
 Werkzeug==0.14.1
diff --git a/subsync/media.py b/subsync/media.py
index 6019fe7..8cc12a5 100644
--- a/subsync/media.py
+++ b/subsync/media.py
@@ -13,6 +13,8 @@
 
 import numpy as np
 import sklearn
+from sklearn import metrics
+
 
 from .ffmpeg import Transcode
 from .log import logger
@@ -150,7 +152,7 @@ def logloss(self, pred, actual, margin=12):
         for i, offset in enumerate(range(-blocks, blocks)):
             snippet = np.roll(actual, offset)
             try:
-                logloss[i] = sklearn.metrics.log_loss(snippet[blocks:-blocks], pred[blocks:-blocks])
+                logloss[i] = metrics.log_loss(snippet[blocks:-blocks], pred[blocks:-blocks])
             except (ValueError, RuntimeWarning):
                 pass
             indices[i] = offset
diff --git a/subsync/net.py b/subsync/net.py
index 76b06e3..bbc870d 100644
--- a/subsync/net.py
+++ b/subsync/net.py
@@ -23,12 +23,13 @@ def summary(self):
 
 
     def load_graph(self, frozen_graph_filename):
-        with tf.gfile.GFile(frozen_graph_filename, "rb") as f:
-            graph_def = tf.GraphDef()
+        # Load the graph with TensorFlow 1.x compatibility mode
+        with tf.compat.v1.gfile.GFile(frozen_graph_filename, "rb") as f:
+            graph_def = tf.compat.v1.GraphDef()
             graph_def.ParseFromString(f.read())
 
         with tf.Graph().as_default() as graph:
-            tf.import_graph_def(
+            tf.compat.v1.import_graph_def(
                 graph_def,
                 input_map=None,
                 return_elements=None,
@@ -40,5 +41,6 @@ def load_graph(self, frozen_graph_filename):
 
     def predict(self, mfcc):
         print("Predicting values...")
-        with tf.Session(graph=self.graph) as sess:
+        with tf.compat.v1.Session(graph=self.graph) as sess:
             return sess.run(self.output, feed_dict={self.input: mfcc})
+