def get_wav_content(files):
for filee in files:
print("Yielding ", filee)
try:
yield tensorflow_wav.get_wav(filee)
except Exception as e:
print("Could not load ", filee, e)
def visualize(sess, dcgan, config, option):
in_wav = tensorflow_wav.get_wav('input.wav')
if option == 0:
z_sample = np.random.uniform(-0.5, 0.5, size=(config.batch_size, dcgan.z_dim))
samples = sess.run(dcgan.sampler, feed_dict={dcgan.z: z_sample})
in_wav['data']=samples
tensorflow_wav.save_wav(in_wav, './samples/test.wav' )
elif option == 1:
values = np.arange(0, 1, 1./config.batch_size)
for idx in range(100):
print(" [*] %d" % idx)
z_sample = np.zeros([config.batch_size, dcgan.z_dim])
for kdx, z in enumerate(z_sample):
z[idx] = values[kdx]
samples = sess.run(dcgan.sampler, feed_dict={dcgan.z: z_sample})
in_wav['data']=samples
tensorflow_wav.save_wav(in_wav, './samples/test_arange_%s.wav'%(idx) )
elif option == 2:
values = np.arange(0, 1, 1./config.batch_size)
for idx in [random.randint(0, 99) for _ in range(100)]:
print(" [*] %d" % idx)
z = np.random.uniform(-0.2, 0.2, size=(dcgan.z_dim))
z_sample = np.tile(z, (config.batch_size, 1))
#z_sample = np.zeros([config.batch_size, dcgan.z_dim])
for kdx, z in enumerate(z_sample):
z[idx] = values[kdx]
samples = sess.run(dcgan.sampler, feed_dict={dcgan.z: z_sample})
make_gif(samples, './samples/test_gif_%s.gif' % (idx))
elif option == 3:
values = np.arange(0, 1, 1./config.batch_size)
for idx in range(100):
print(" [*] %d" % idx)
z_sample = np.zeros([config.batch_size, dcgan.z_dim])
for kdx, z in enumerate(z_sample):
z[idx] = values[kdx]
samples = sess.run(dcgan.sampler, feed_dict={dcgan.z: z_sample})
make_gif(samples, './samples/test_gif_%s.gif' % (idx))
elif option == 4:
wav_set = []
values = np.arange(0, 1, 1./config.batch_size)
for idx in range(100):
print(" [*] %d" % idx)
z_sample = np.zeros([config.batch_size, dcgan.z_dim])
for kdx, z in enumerate(z_sample): z[idx] = values[kdx]
wav_set.append(sess.run(dcgan.sampler, feed_dict={dcgan.z: z_sample}))
make_gif(wav_set[-1], './samples/test_gif_%s.gif' % (idx))
new_wav_set = [merge(np.array([wavs[idx] for wavs in wav_set]), [10, 10]) \
for idx in range(64) + range(63, -1, -1)]
make_gif(new_wav_set, './samples/test_gif_merged.gif', duration=8)
def test_sftf1(self):
wav_path="input.wav"
fs=2048
T=1
framesz=(64/2048)
hop=(T*2048-64)/(2048*64)
wav= tensorflow_wav.get_wav(wav_path)
data = wav['data'][:fs*T]
with tf.Session() as sess:
x = test_stft_cpu.stft(data, fs, framesz, hop)
print('max/min', x.max(),x.min())
raw_data = tf.placeholder(tf.complex64, [fs*T])
tstft=tensorflow_wav.stft(raw_data, fs, framesz, hop)
y = sess.run(tstft, {raw_data:data})
print(x)
print(y)
self.assertEqual(True, True)
def test_sftf1(self):
wav_path = "input.wav"
fs = 2048
T = 1
framesz = (64 / 2048)
hop = (T * 2048 - 64) / (2048 * 64)
wav = tensorflow_wav.get_wav(wav_path)
data = wav['data'][:fs * T]
with tf.Session() as sess:
x = test_stft_cpu.stft(data, fs, framesz, hop)
print('max/min', x.max(), x.min())
raw_data = tf.placeholder(tf.complex64, [fs * T])
tstft = tensorflow_wav.stft(raw_data, fs, framesz, hop)
y = sess.run(tstft, {raw_data: data})
print(x)
print(y)
self.assertEqual(True, True)
def test_isftf1(self):
wav_path="input.wav"
fs=2048
T=1
framesz=(64/2048)
hop=(T*2048-64)/(2048*64)
wav= tensorflow_wav.get_wav(wav_path)
data = wav['data'][:fs*T]
with tf.Session() as sess:
xin = test_stft_cpu.stft(data, fs, framesz, hop)
print("Xin is", xin)
raw_data = tf.placeholder(tf.complex64, [64, 64])
x=test_stft_cpu.istft(xin, fs, T, hop)
tstft=tensorflow_wav.istft(raw_data, fs, hop)
y = sess.run(tstft, {raw_data:xin})
print('x is', x)
print('y is', y)
self.assertEqual(x.tolist()[0:10], y.tolist()[0:10])
def test_isftf1(self):
wav_path = "input.wav"
fs = 2048
T = 1
framesz = (64 / 2048)
hop = (T * 2048 - 64) / (2048 * 64)
wav = tensorflow_wav.get_wav(wav_path)
data = wav['data'][:fs * T]
with tf.Session() as sess:
xin = test_stft_cpu.stft(data, fs, framesz, hop)
print("Xin is", xin)
raw_data = tf.placeholder(tf.complex64, [64, 64])
x = test_stft_cpu.istft(xin, fs, T, hop)
tstft = tensorflow_wav.istft(raw_data, fs, hop)
y = sess.run(tstft, {raw_data: xin})
print('x is', x)
print('y is', y)
self.assertEqual(x.tolist()[0:10], y.tolist()[0:10])
framesamp = X.shape[1]
print('fsamp',framesamp)
hopsamp = int(hop*fs)
for n,i in enumerate(range(0, len(x)-framesamp, hopsamp)):
if(n>=X.shape[0]):
break
#print("setting i to i+framesamp", n, i, framesamp, i+framesamp, len(X), len(x))
x[i:i+framesamp] += scipy.real(ifft(X[n]))
#print(x.shape)
return x
if __name__ == '__main__':
wav_path="input.wav"
wav= tensorflow_wav.get_wav(wav_path)
fs=2048
T=1
data = wav['data'][:fs*T]
print("data, max, mean, stddev",data.min(), data.max(), np.mean(data), np.std(data))
#data['sampwidth']
framesz=(64./2048)
print("fs is ", fs)
print("Framesz is", framesz)
hop=1./64
print("Hop is", hop)
framesamp = int(framesz*fs)
hopsamp = int(hop*fs)
print("framesamp is", framesamp)
print("hopsamp is", hopsamp, hop, fs)
print('size is',data-framesamp, 'stride', hopsamp)
# should be [w, h, z] where w=441, h=?(0.25 seconds?), and z=2(complex, real)
# Sanity test load and save
import tensorflow as tf
import numpy as np
from numpy.fft import fft, ifft
from tensorflow_wav import get_wav, save_wav, encode, decode
wav_path="input.wav"
wav_size=64
with tf.Session() as sess:
wav= get_wav(wav_path)
print('data is', wav['data'])
print(wav)
fs = 2048
T=1
raw_data = tf.placeholder(tf.complex64, [1,fs*T])
#print("WAV IS", wav['data'].tolist())
data = wav['data'][:int(fs*T)]
data = data.reshape([1,int(fs*T)])
#data = tf.reshape(raw_data[:64*64*64], [-1])
encoded = encode(raw_data, bitrate=fs)
#print(encoded)
decoded = decode(encoded, bitrate=fs)
noop = decoded
import os
from glob import glob
import tensorflow as tf
import numpy as np
import tensorflow_wav
from model import CPPNVAE
model = CPPNVAE(batch_size=1)
model.load_model('save')
print("Loaded")
data = glob(os.path.join("./training", "*.wav"))
sample_file = data[0]
sample = tensorflow_wav.get_wav(sample_file)
def new_z():
return np.random.normal(size=(1, 32)).astype(np.float32)
print("Generating")
bitrate = 4096
t_dims = [bitrate, bitrate * 2]
scales = [8.0, 16.0, 24.0]
for t_dim in t_dims:
for scale in scales:
wavdata = [
model.generate(new_z(), t_dim=t_dim, scale=scale)
for i in range(10)
]
fs = 2048
framesz = (64 / 2048)
hop = (2048 - 64) / (2048 * 64)
import glob
from test_stft_cpu import stft
import os
import sys
import tensorflow_wav
import numpy as np
import math
if (len(sys.argv) < 2):
print("You have to pick a file")
no_file_Picked_Exception
files = glob.glob(sys.argv[1])
if (len(files) == 0):
print("Who you trying to kid?")
for file in files:
print("Reading stft for " + file, ' at rate ', fs)
wav = tensorflow_wav.get_wav(file)
data = stft(wav['data'], fs, framesz, hop)
print(wav)
wav['data'] = data.real
print(wav['data'])
#wav['data'] = np.sign(wav['data'])*np.power(wav['data'], 2)
print(np.min(wav['data']), np.max(wav['data']))
res = tensorflow_wav.save_stft(wav, file + ".stft")
print(file + ".stft" + " is written")