from scipy.io import wavfile import os import sys from kdllib import load_checkpoint, theano_one_hot, concatenate from kdllib import fetch_fruitspeech_spectrogram, list_iterator from kdllib import np_zeros, GRU, GRUFork, dense_to_one_hot from kdllib import make_weights, make_biases, relu, run_loop from kdllib import as_shared, adam, gradient_clipping from kdllib import get_values_from_function, set_shared_variables_in_function from kdllib import soundsc, categorical_crossentropy from kdllib import sample_binomial, sigmoid if __name__ == "__main__": import argparse speech = fetch_fruitspeech_spectrogram() X = speech["data"] y = speech["target"] vocabulary = speech["vocabulary"] vocabulary_size = speech["vocabulary_size"] reconstruct = speech["reconstruct"] fs = speech["sample_rate"] X = np.array([x.astype(theano.config.floatX) for x in X]) y = np.array([yy.astype(theano.config.floatX) for yy in y]) minibatch_size = 1 n_epochs = 200 # Used way at the bottom in the training loop! checkpoint_every_n = 10 cut_len = 41 # Used way at the bottom in the training loop! random_state = np.random.RandomState(1999)
from theano.sandbox.rng_mrg import MRG_RandomStreams as RandomStreams from scipy.io import wavfile import os import sys from kdllib import make_conv_weights, conv2d_transpose, unpool, softmax, make_biases from kdllib import load_checkpoint, dense_to_one_hot, categorical_crossentropy from kdllib import fetch_fruitspeech_spectrogram, list_iterator, np_zeros, GRU, GRUFork from kdllib import make_weights, as_shared, adam, gradient_clipping, theano_one_hot from kdllib import get_values_from_function, set_shared_variables_in_function from kdllib import save_checkpoint, save_weights, relu, tanh, soundsc from kdllib import run_loop if __name__ == "__main__": import argparse speech = fetch_fruitspeech_spectrogram() X = speech["data"] y = speech["target"] vocabulary = speech["vocabulary"] vocabulary_size = speech["vocabulary_size"] reconstruct = speech["reconstruct"] fs = speech["sample_rate"] X = np.array([x.astype(theano.config.floatX) for x in X]) y = np.array([yy.astype(theano.config.floatX) for yy in y]) minibatch_size = 20 n_epochs = 20000 # Used way at the bottom in the training loop! checkpoint_every_n = 500 # Was 300 for handwriting cut_len = 31 # Used way at the bottom in the training loop! random_state = np.random.RandomState(1999)