-
Notifications
You must be signed in to change notification settings - Fork 0
/
data.py
68 lines (58 loc) · 2.71 KB
/
data.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
import pandas as pd
import numpy as np
import librosa
import tensorflow as tf
from sklearn.model_selection import train_test_split
# Creating labels
def create_labels():
bird_labels = pd.read_csv("dataset/bird_songs_metadata.csv", usecols=['species'])
bird_labels = bird_labels.values
bird_labels[bird_labels == ['bewickii']] = 0
bird_labels[bird_labels == ['polyglottos']] = 1
bird_labels[bird_labels == ['migratorius']] = 2
bird_labels[bird_labels == ['melodia']] = 3
bird_labels[bird_labels == ['cardinalis']] = 4
bird_labels = np.squeeze(bird_labels)
return bird_labels
# Creating list of file paths
def create_filepaths():
file_names = pd.read_csv("dataset/bird_songs_metadata.csv", usecols=['filename'])
file_names = np.squeeze(file_names.values)
bird_filepaths = np.array([])
for file in file_names:
bird_filepaths = np.append(bird_filepaths, 'dataset/wavfiles/'+file)
return bird_filepaths
# Splitting data into training and validation set
def split_data():
bird_filepaths = create_filepaths()
bird_labels = create_labels()
bird_filepaths_train, bird_filepaths_val, bird_labels_train, bird_labels_val = train_test_split(
bird_filepaths, bird_labels, test_size=0.10, random_state=2419)
return bird_filepaths_train, bird_filepaths_val, bird_labels_train, bird_labels_val
# Function that reads filepaths
def read_file(path):
y,_ = librosa.load(path)
return y
# Converting to db scale
def spec_to_db(y):
y_db = librosa.amplitude_to_db(y, ref=100)
return y_db
# Map function that returns spectrograms
def map_function(path_tensor, label):
y=tf.numpy_function(read_file, inp= [path_tensor], Tout = tf.float32)
spectrogram = tf.abs(tf.signal.stft(y, frame_length=512, frame_step=64))
spectrogram_db = tf.numpy_function(spec_to_db, inp = [spectrogram], Tout = tf.float32)
spectrogram_db = spectrogram_db/80+1
return spectrogram_db, label
# Function that creates tf.data.Dataset with shuffle, repeat, map and batch
def make_dataset(bird_labels, bird_filepaths, shuffle):
bird_labels = tf.convert_to_tensor(bird_labels, dtype = tf.int32)
bird_filepaths = tf.convert_to_tensor(bird_filepaths, dtype = tf.string)
bird_labels = tf.data.Dataset.from_tensor_slices(bird_labels)
bird_filepaths = tf.data.Dataset.from_tensor_slices(bird_filepaths)
dataset = tf.data.Dataset.zip( bird_filepaths, bird_labels)
if shuffle:
dataset = dataset.shuffle(buffer_size = dataset.cardinality(), reshuffle_each_iteration=True)
dataset = dataset.map(map_function, num_parallel_calls = tf.data.AUTOTUNE)
dataset = dataset.batch(batch_size = 32, num_parallel_calls = tf.data.AUTOTUNE, drop_remainder = True)
return dataset