From: Andrew Gundersen <53452154+gundersena@users.noreply.github.com>
Date: Mon, 17 Feb 2020 22:48:01 +0000 (-0600)
Subject: Add data folder to repo
X-Git-Url: https://andrewgundersen.net/repos?a=commitdiff_plain;h=cfdab59a341833af0a91821ca8ae1945ce69376a;p=ml-audio-enhancer

Add data folder to repo
---

diff --git a/data/make_text.py b/data/make_text.py
new file mode 100644
index 0000000..4932d1b
--- /dev/null
+++ b/data/make_text.py
@@ -0,0 +1,29 @@
+import os
+import random
+import librosa
+
+file_path = 'VCTK-Corpus/wav48'
+dataset_path = './multispeaker'
+
+folders = os.listdir(file_path)
+
+for folder in folders:
+    if folder == '.DS_Store':
+        folders.remove(folder)
+print (folders)
+
+for folder in folders:
+    files = os.listdir(os.path.join(file_path, folder))
+    random.shuffle(files)
+    for file in files:
+        if len(x) < 10:
+            input('This file is corrupted')
+        u = random.uniform(0,1) # a single value is returned between 0 and 1
+        if u > 0.1:
+            with open(f'{dataset_path}/train-files1.txt', 'a+') as text_file:
+                text_file.write(f'{file_path}/{folder}/{file}')
+                text_file.write('\n')
+        if u < 0.1:
+            with open(f'{dataset_path}/val-files1.txt', 'a+') as text_file:
+                text_file.write(f'{file_path}/{folder}/{file}')
+                text_file.write('\n')
diff --git a/data/multispeaker/test_data.py b/data/multispeaker/test_data.py
new file mode 100644
index 0000000..497dcd0
--- /dev/null
+++ b/data/multispeaker/test_data.py
@@ -0,0 +1,48 @@
+# import a dataset and write spectrograms
+from matplotlib import pyplot as plt
+import numpy as np
+import librosa
+import h5py
+
+
+def save_spectrum(S_lr, S_ir, S_pr, outfile, lim=1000):
+    plt.subplot(1,3,1)
+    plt.title('Data')
+    plt.xlabel('Frequency')
+    plt.ylabel('Time')
+    plt.imshow(S_lr, aspect=10)
+
+    plt.subplot(1,3,2)
+    plt.title('Label')
+    plt.xlabel('Frequency')
+    plt.ylabel('Time')
+    plt.imshow(S_ir, aspect=10)
+
+    plt.subplot(1,3,3)
+    plt.title('Label')
+    plt.xlabel('Frequency')
+    plt.ylabel('Time')
+    plt.imshow(S_pr, aspect=10)
+
+    plt.tight_layout()
+    plt.savefig(outfile)
+
+
+def get_spectrum(x, n_fft=2048):
+    S = librosa.stft(x, n_fft)
+    S = np.log1p(np.abs(S))
+    p = np.angle(S)
+    S = np.log1p(np.abs(S))
+    return S.T
+
+
+file = 'vctk-train.4.16000.800.10.0.25.h5'
+
+with h5py.File(file, 'r') as hf:
+    X = np.array(hf.get('data'))
+    Y = np.array(hf.get('label'))
+    print(X)
+
+data = get_spectrum(X.flatten(), n_fft=2048)
+label = get_spectrum(Y.flatten(), n_fft=2048)
+save_spectrum(data, label, outfile='test1.png')