In [1]:

    

import tensorflow as tf
import os
import matplotlib.pyplot as plt
import numpy as np


    

In [2]:

    

os.environ.setdefault('GLUE_SETTINGS_MODULE', 'settings')


    

    
Out[2]:





'settings'

In [3]:

    

import jbb.model1 as nw
import settings16 as settings


    

In [4]:

    

batch_size = settings.BATCH_SIZE
samplerate = settings.SAMPLERATE


    

In [5]:

    

x = tf.placeholder(tf.float32, [1, samplerate])
cls = tf.placeholder(tf.int32, 1)


    

In [6]:

    

with tf.variable_scope('network'):
    net = nw.conv_net(x, cls, None)


    

    




conv1:  Tensor("network/glu_conv1_1/mul:0", shape=(1, 16384, 1, 4), dtype=float32)
conv1:  Tensor("network/glu_conv1_2/mul:0", shape=(1, 16384, 1, 4), dtype=float32)
conv2:  Tensor("network/glu_conv2_1/mul:0", shape=(1, 16384, 1, 8), dtype=float32)
conv2:  Tensor("network/glu_conv2_2/mul:0", shape=(1, 16384, 1, 8), dtype=float32)
conv2:  Tensor("network/MaxPool2D/MaxPool:0", shape=(1, 8192, 1, 8), dtype=float32)
embedded: Tensor("network/embedding/concat:0", shape=(1, 8192, 1, 10), dtype=float32)
conv3:  Tensor("network/glu_conv3_2/mul:0", shape=(1, 8192, 1, 16), dtype=float32)
d2sp:  Tensor("network/d2s1/Reshape:0", shape=(1, 16384, 1, 8), dtype=float32)
conv4:  Tensor("network/glu_conv4_3/mul:0", shape=(1, 16384, 1, 64), dtype=float32)
conv5:  Tensor("network/conv5/Relu:0", shape=(1, 16384, 1, 256), dtype=float32)
out:  Tensor("network/output/Reshape:0", shape=(1, 16384, 256), dtype=float32)

In [7]:

    

saver = tf.train.Saver()


    

In [8]:

    

sess = tf.InteractiveSession()


    

In [9]:

    

saver.restore(sess, tf.train.latest_checkpoint('logs','emb_mid_r3.checkpoint'))


    

    




INFO:tensorflow:Restoring parameters from /home/s123094/FlowFairy/examples/sine_fix/logs/emb_mid_r3.ckpt-113000

In [10]:

    

coll = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES)
c1uw = coll[0]


    

In [11]:

    

i=0
for i, v in enumerate(coll):
    print(i,v)


    

    




0 <tf.Variable 'network/conv1_1_unit/weights:0' shape=(256, 1, 1, 4) dtype=float32_ref>
1 <tf.Variable 'network/conv1_1_unit/BatchNorm/beta:0' shape=(4,) dtype=float32_ref>
2 <tf.Variable 'network/conv1_1_gate/weights:0' shape=(256, 1, 1, 4) dtype=float32_ref>
3 <tf.Variable 'network/conv1_1_gate/BatchNorm/beta:0' shape=(4,) dtype=float32_ref>
4 <tf.Variable 'network/conv1_2_unit/weights:0' shape=(256, 1, 4, 4) dtype=float32_ref>
5 <tf.Variable 'network/conv1_2_unit/biases:0' shape=(4,) dtype=float32_ref>
6 <tf.Variable 'network/conv1_2_gate/weights:0' shape=(256, 1, 4, 4) dtype=float32_ref>
7 <tf.Variable 'network/conv1_2_gate/biases:0' shape=(4,) dtype=float32_ref>
8 <tf.Variable 'network/conv2_1_unit/weights:0' shape=(128, 1, 4, 8) dtype=float32_ref>
9 <tf.Variable 'network/conv2_1_unit/biases:0' shape=(8,) dtype=float32_ref>
10 <tf.Variable 'network/conv2_1_gate/weights:0' shape=(128, 1, 4, 8) dtype=float32_ref>
11 <tf.Variable 'network/conv2_1_gate/biases:0' shape=(8,) dtype=float32_ref>
12 <tf.Variable 'network/conv2_2_unit/weights:0' shape=(128, 1, 8, 8) dtype=float32_ref>
13 <tf.Variable 'network/conv2_2_unit/biases:0' shape=(8,) dtype=float32_ref>
14 <tf.Variable 'network/conv2_2_gate/weights:0' shape=(128, 1, 8, 8) dtype=float32_ref>
15 <tf.Variable 'network/conv2_2_gate/biases:0' shape=(8,) dtype=float32_ref>
16 <tf.Variable 'network/embedding/embedding:0' shape=(200, 2) dtype=float32_ref>
17 <tf.Variable 'network/conv3_1_unit/weights:0' shape=(128, 1, 10, 16) dtype=float32_ref>
18 <tf.Variable 'network/conv3_1_unit/biases:0' shape=(16,) dtype=float32_ref>
19 <tf.Variable 'network/conv3_1_gate/weights:0' shape=(128, 1, 10, 16) dtype=float32_ref>
20 <tf.Variable 'network/conv3_1_gate/biases:0' shape=(16,) dtype=float32_ref>
21 <tf.Variable 'network/conv3_2_unit/weights:0' shape=(128, 1, 16, 16) dtype=float32_ref>
22 <tf.Variable 'network/conv3_2_unit/biases:0' shape=(16,) dtype=float32_ref>
23 <tf.Variable 'network/conv3_2_gate/weights:0' shape=(128, 1, 16, 16) dtype=float32_ref>
24 <tf.Variable 'network/conv3_2_gate/biases:0' shape=(16,) dtype=float32_ref>
25 <tf.Variable 'network/conv4_1_unit/weights:0' shape=(128, 1, 8, 16) dtype=float32_ref>
26 <tf.Variable 'network/conv4_1_unit/biases:0' shape=(16,) dtype=float32_ref>
27 <tf.Variable 'network/conv4_1_gate/weights:0' shape=(128, 1, 8, 16) dtype=float32_ref>
28 <tf.Variable 'network/conv4_1_gate/biases:0' shape=(16,) dtype=float32_ref>
29 <tf.Variable 'network/conv4_2_unit/weights:0' shape=(128, 1, 20, 32) dtype=float32_ref>
30 <tf.Variable 'network/conv4_2_unit/biases:0' shape=(32,) dtype=float32_ref>
31 <tf.Variable 'network/conv4_2_gate/weights:0' shape=(128, 1, 20, 32) dtype=float32_ref>
32 <tf.Variable 'network/conv4_2_gate/biases:0' shape=(32,) dtype=float32_ref>
33 <tf.Variable 'network/conv4_3_unit/weights:0' shape=(128, 1, 32, 64) dtype=float32_ref>
34 <tf.Variable 'network/conv4_3_unit/biases:0' shape=(64,) dtype=float32_ref>
35 <tf.Variable 'network/conv4_3_gate/weights:0' shape=(128, 1, 32, 64) dtype=float32_ref>
36 <tf.Variable 'network/conv4_3_gate/biases:0' shape=(64,) dtype=float32_ref>
37 <tf.Variable 'network/conv5/weights:0' shape=(1, 1, 64, 256) dtype=float32_ref>
38 <tf.Variable 'network/conv5/biases:0' shape=(256,) dtype=float32_ref>

In [12]:

    

embedding = coll[16]
print(embedding)
plt.plot(embedding.eval()[:,0])
plt.plot(embedding.eval()[:,1])
plt.plot(embedding.eval().sum(1))
plt.show()


    

    




<tf.Variable 'network/embedding/embedding:0' shape=(200, 2) dtype=float32_ref>






    

In [13]:

    

def classify(val):
    val = (val-np.min(val))/(np.max(val)-np.min(val))
    return np.floor(val*255)

def getid(frq):
    limits = settings.FREQUENCY_LIMIT
    count = settings.CLASS_COUNT
    step = (limits[1] - limits[0]) // count
    return (frq-limits[0])//step

def getfrq(fid):
    limits = settings.FREQUENCY_LIMIT
    count = settings.CLASS_COUNT
    step = (limits[1] - limits[0]) // count
    return limits[0] + step*fid


    

In [14]:

    

with tf.variable_scope('network', reuse=True):
    tnet = nw.conv_net(x, cls, None, is_training=True)


    

    




conv1:  Tensor("network_1/glu_conv1_1/mul:0", shape=(1, 16384, 1, 4), dtype=float32)
conv1:  Tensor("network_1/glu_conv1_2/mul:0", shape=(1, 16384, 1, 4), dtype=float32)
conv2:  Tensor("network_1/glu_conv2_1/mul:0", shape=(1, 16384, 1, 8), dtype=float32)
conv2:  Tensor("network_1/glu_conv2_2/mul:0", shape=(1, 16384, 1, 8), dtype=float32)
conv2:  Tensor("network_1/MaxPool2D/MaxPool:0", shape=(1, 8192, 1, 8), dtype=float32)
embedded: Tensor("network_1/embedding/concat:0", shape=(1, 8192, 1, 10), dtype=float32)
conv3:  Tensor("network_1/glu_conv3_2/mul:0", shape=(1, 8192, 1, 16), dtype=float32)
d2sp:  Tensor("network_1/d2s1/Reshape:0", shape=(1, 16384, 1, 8), dtype=float32)
conv4:  Tensor("network_1/glu_conv4_3/mul:0", shape=(1, 16384, 1, 64), dtype=float32)
conv5:  Tensor("network_1/conv5/Relu:0", shape=(1, 16384, 1, 256), dtype=float32)
out:  Tensor("network_1/output/Reshape:0", shape=(1, 16384, 256), dtype=float32)

In [15]:

    

pred = tf.argmax(net, 2)


    

In [16]:

    

tpred = tf.argmax(tnet, 2)


    

In [30]:

    

frqid = 10
frq = getfrq(frqid)
print(frq)

bfrq = getfrq(100)
print(bfrq)

myrange = np.arange(2**14) * 2 * np.pi / 2**14
mysine = np.sin(myrange * frq)
mysine2 = np.sin(myrange * bfrq)
noise = np.random.uniform(-0.5, 0.5, samplerate)

#plt.plot(mysine)
#plt.show()


    

    




320
500

In [31]:

    

b = np.array([mysine+mysine2+noise])
y = np.array([frqid])
run = sess.run(pred, feed_dict={x: b, cls: y})


    

In [32]:

    

#plt.plot(classify(b[0][1000:2000]))
plt.plot(run[0][1000:1200], 'r')
plt.plot(classify(mysine)[1000:1200], 'b', alpha=0.5)
plt.plot(classify(b[0][1000:1200]), 'g', alpha=0.5)
plt.show()


    

In [20]:

    

(150-120)/3


    

    
Out[20]:





10.0

In [ ]: