diff --git a/speech_data.py b/speech_data.py
index 66b6d10..a5c166e 100644
--- a/speech_data.py
+++ b/speech_data.py
@@ -31,11 +31,12 @@ def create_test_pair(l, r, max_samples):
     r_sample = append_zeros(r, max_samples)
     return np.asarray([[l_sample, r_sample]])
 
+
 def create_X(sp, max_samples):
-    return create_pair(sp[0]['spectrogram'],sp[1]['spectrogram'],max_samples)
+    return create_pair(sp[0]['spectrogram'], sp[1]['spectrogram'], max_samples)
 
 
-def get_word_pairs_data(word,max_samples):
+def get_word_pairs_data(word, max_samples):
     audio_samples = pd.read_csv(
         './outputs/audio.csv',
         names=['word', 'voice', 'rate', 'variant', 'file'])
@@ -55,7 +56,7 @@ def get_word_pairs_data(word,max_samples):
     Y = np.hstack([np.ones(len(same_data)), np.zeros(len(diff_data))])
     X = np.asarray(same_data + diff_data)
     # tr_pairs, te_pairs, tr_y, te_y = train_test_split(X, Y, test_size=0.1)
-    return (X,Y)
+    return (X, Y)
 
 
 def create_spectrogram_data(audio_group='audio'):
@@ -106,8 +107,8 @@ def create_speech_pairs_data(audio_group='audio'):
 def speech_model_data():
     tr_pairs = np.load('outputs/tr_pairs.npy') / 255.0
     te_pairs = np.load('outputs/te_pairs.npy') / 255.0
-    tr_pairs[tr_pairs < 0] = 0
-    te_pairs[te_pairs < 0] = 0
+    # tr_pairs[tr_pairs < 0] = 0
+    # te_pairs[te_pairs < 0] = 0
     tr_y = np.load('outputs/tr_y.npy')
     te_y = np.load('outputs/te_y.npy')
     return tr_pairs, te_pairs, tr_y, te_y
diff --git a/test_siamese.py b/test_siamese.py
index a83f478..af66f79 100644
--- a/test_siamese.py
+++ b/test_siamese.py
@@ -6,12 +6,17 @@ reload(speech_data)
 from speech_data import create_test_pair,get_word_pairs_data
 import numpy as np
 
-sunflower_data,sunflower_result = get_word_pairs_data('sunflowers',15)
-sunflower_result
 model = siamese_model((15, 1654))
 model.load_weights('./models/siamese_speech_model-final.h5')
-spec1 = record_spectrogram(n_sec=1.4)
-spec2 = record_spectrogram(n_sec=1.4)
-inp = create_test_pair(spec1,spec2,16)
-model.predict([inp[:, 0], inp[:, 1]])
+
+def predict_recording_with(m,sample_size=15):
+    spec1 = record_spectrogram(n_sec=1.4)
+    spec2 = record_spectrogram(n_sec=1.4)
+    inp = create_test_pair(spec1,spec2,sample_size)
+    return m.predict([inp[:, 0], inp[:, 1]])
+
+predict_recording_with(model)
+
+sunflower_data,sunflower_result = get_word_pairs_data('sunflowers',15)
+sunflower_result
 model.predict([sunflower_data[:, 0], sunflower_data[:, 1]]) < 0.5