def get_config(batch_size, data_path):
return configs.Config(
model=MusicVAE(lstm_models.BidirectionalLstmEncoder(),
lstm_models.CategoricalLstmDecoder()),
hparams=merge_hparams(
lstm_models.get_default_hparams(),
HParams(
batch_size=512,
max_seq_len=32, # 2 bars w/ 16 steps per bar
z_size=512,
enc_rnn_size=[2048],
dec_rnn_size=[2048, 2048, 2048],
free_bits=0,
max_beta=0.5,
beta_rate=0.99999,
sampling_schedule='inverse_sigmoid',
sampling_rate=1000,
)),
note_sequence_augmenter=data.NoteSequenceAugmenter(
transpose_range=(-5, 5)),
data_converter=data.OneHotMelodyConverter(
valid_programs=data.MEL_PROGRAMS,
skip_polyphony=False,
max_bars=100, # Truncate long melodies before slicing.
slice_bars=2,
steps_per_quarter=4),
train_examples_path=data_path,
eval_examples_path=data_path,
)
def _update_config(config1, config2):
"""Update config1 hparams with hparams from config2"""
h = merge_hparams(config1.hparams, config2.hparams)
return Config(model=config1.model,
hparams=h,
note_sequence_augmenter=config1.note_sequence_augmenter,
data_converter=config1.data_converter,
train_examples_path=config1.train_examples_path,
eval_examples_path=config1.eval_examples_path)
return Config(**config_dict)
CONFIG_MAP = {}
# Melody
CONFIG_MAP['cat-mel_2bar_small'] = Config(
model=MusicVAE(lstm_models.BidirectionalLstmEncoder(),
lstm_models.CategoricalLstmDecoder()),
hparams=merge_hparams(
lstm_models.get_default_hparams(),
HParams(
batch_size=512,
max_seq_len=32, # 2 bars w/ 16 steps per bar
z_size=256,
enc_rnn_size=[512],
dec_rnn_size=[256, 256],
free_bits=0,
max_beta=0.2,
beta_rate=0.99999,
sampling_schedule='inverse_sigmoid',
sampling_rate=1000,
)),
note_sequence_augmenter=data.NoteSequenceAugmenter(transpose_range=(-5,
5)),
data_converter=data.OneHotMelodyConverter(
valid_programs=data.MEL_PROGRAMS,
skip_polyphony=False,
max_bars=100, # Truncate long melodies before slicing.
slice_bars=2,
steps_per_quarter=4),
train_examples_path=None,
config_dict = config.values()
config_dict.update(update_dict)
return Config(**config_dict)
config_map = {}
# Melody
config_map['cat-mel_2bar_small'] = Config(
model=MusicVAE(lstm_models.BidirectionalLstmEncoder(),
lstm_models.CategoricalLstmDecoder()),
hparams=merge_hparams(
lstm_models.get_default_hparams(),
HParams(
batch_size=512,
max_seq_len=32, # 2 bars w/ 16 steps per bar
z_size=256,
enc_rnn_size=[512],
dec_rnn_size=[256, 256],
)),
note_sequence_augmenter=None,
note_sequence_converter=data.OneHotMelodyConverter(
valid_programs=data.MEL_PROGRAMS,
skip_polyphony=True,
max_bars=100, # Truncate long melodies before slicing.
slice_bars=2,
steps_per_quarter=4),
train_examples_path=None,
eval_examples_path=None,
)
config_dict.update(update_dict)
return Config(**config_dict)
CONFIG_MAP = {}
# Melody
CONFIG_MAP['cat-mel_2bar_small'] = Config(
model=MusicVAE(lstm_models.BidirectionalLstmEncoder(),
lstm_models.CategoricalLstmDecoder()),
hparams=merge_hparams(
lstm_models.get_default_hparams(),
HParams(
batch_size=512,
max_seq_len=32, # 2 bars w/ 16 steps per bar
z_size=256,
enc_rnn_size=[512],
dec_rnn_size=[256, 256],
)),
note_sequence_augmenter=data.NoteSequenceAugmenter(transpose_range=(-5, 5)),
data_converter=data.OneHotMelodyConverter(
valid_programs=data.MEL_PROGRAMS,
skip_polyphony=False,
max_bars=100, # Truncate long melodies before slicing.
slice_bars=2,
steps_per_quarter=4),
train_examples_path=None,
eval_examples_path=None,
)
config_dict.update(update_dict)
return Config(**config_dict)
CONFIG_MAP = {}
# Melody
CONFIG_MAP['cat-mel_2bar_small'] = Config(
model=MusicVAE(lstm_models.BidirectionalLstmEncoder(),
lstm_models.CategoricalLstmDecoder()),
hparams=merge_hparams(
lstm_models.get_default_hparams(),
HParams(
batch_size=512,
max_seq_len=32, # 2 bars w/ 16 steps per bar
z_size=256,
enc_rnn_size=[512],
dec_rnn_size=[256, 256],
)),
note_sequence_augmenter=None,
note_sequence_converter=data.OneHotMelodyConverter(
valid_programs=data.MEL_PROGRAMS,
skip_polyphony=False,
max_bars=100, # Truncate long melodies before slicing.
slice_bars=2,
steps_per_quarter=4),
train_examples_path=None,
eval_examples_path=None,
)
CONFIG_MAP = {}
# Melody
CONFIG_MAP['cat-mel_2bar_small'] = Config(
model=MusicVAE(lstm_models.BidirectionalLstmEncoder(),
lstm_models.CategoricalLstmDecoder()),
hparams=merge_hparams(
lstm_models.get_default_hparams(),
HParams(
batch_size=512,
max_seq_len=32, # 2 bars w/ 16 steps per bar
z_size=256,
enc_rnn_size=[512],
dec_rnn_size=[256, 256],
free_bits=0,
max_beta=0.2,
beta_rate=0.99999,
sampling_schedule='inverse_sigmoid',
sampling_rate=1000,
)),
note_sequence_augmenter=data.NoteSequenceAugmenter(transpose_range=(-5, 5)),
data_converter=data.OneHotMelodyConverter(
valid_programs=data.MEL_PROGRAMS,
skip_polyphony=False,
max_bars=100, # Truncate long melodies before slicing.
slice_bars=2,
steps_per_quarter=4),
train_examples_path=None,
eval_examples_path=None,
from magenta.models.music_vae.data import BASS_PROGRAMS
from magenta.models.music_vae.data import NoteSequenceAugmenter
from magenta.models.music_vae.data import OneHotMelodyConverter
from magenta.models.music_vae.music_vae_train import FLAGS
from magenta.models.music_vae.music_vae_train import run
CONFIG_MAP["cat-bass_2bar_small"] = Config(
model=MusicVAE(lstm_models.BidirectionalLstmEncoder(),
lstm_models.CategoricalLstmDecoder()),
hparams=merge_hparams(
lstm_models.get_default_hparams(),
tf.contrib.training.HParams(
batch_size=512,
max_seq_len=32,
z_size=256,
enc_rnn_size=[512],
dec_rnn_size=[256, 256],
free_bits=0,
max_beta=0.2,
beta_rate=0.99999,
sampling_schedule="inverse_sigmoid",
sampling_rate=1000,
)),
note_sequence_augmenter=NoteSequenceAugmenter(transpose_range=(-5, 5)),
data_converter=OneHotMelodyConverter(
valid_programs=BASS_PROGRAMS,
skip_polyphony=False,
max_bars=100,
slice_bars=2,
steps_per_quarter=4),
train_examples_path=None,
eval_examples_path=None,
config_dict.update(update_dict) return Config(**config_dict) CONFIG_MAP = dict() CONFIG_MAP['lc-cat-mel_2bar_big'] = Config( model=LCMusicVAE(lstm_models.BidirectionalLstmEncoder(), lstm_models.CategoricalLstmDecoder()), hparams=merge_hparams( lstm_models.get_default_hparams(), HParams( batch_size=2, max_seq_len=32, # 2 bars w/ 16 steps per bar z_size=512, encoded_z_size=8, enc_rnn_size=[2048], dec_rnn_size=[128, 128], free_bits=0, max_beta=0.5, beta_rate=0.99999, sampling_schedule='inverse_sigmoid', sampling_rate=1000, )), note_sequence_augmenter=data.NoteSequenceAugmenter(transpose_range=(-5, 5)), data_converter=data.OneHotMelodyConverter( valid_programs=data.MEL_PROGRAMS, skip_polyphony=False, max_bars=100, # Truncate long melodies before slicing. slice_bars=2, steps_per_quarter=4), train_examples_path=None,
gap_bars=2)
# simplest version of the 16 bar trio because it has the largest size
# and I can understand it best
CONFIG_MAP['flat-trio_16bar'] = Config(
model=MusicVAE(
lstm_models.BidirectionalLstmEncoder(),
lstm_models.MultiOutCategoricalLstmDecoder(output_depths=[
90, # melody
90, # bass
512, # drums
])),
hparams=merge_hparams(
lstm_models.get_default_hparams(),
HParams(
batch_size=256,
max_seq_len=256,
z_size=512,
enc_rnn_size=[2048, 2048],
dec_rnn_size=[2048, 2048, 2048],
)),
note_sequence_augmenter=None,
data_converter=trio_16bar_converter,
train_examples_path='./MidiSet1/',
eval_examples_path=None,
)
def main():
# test1()
import_songs()
slice_bars=16,
steps_per_quarter=4)
CONFIG_MAP['16bar_44_1'] = Config(
model=MusicVAE(
lstm_models.BidirectionalLstmEncoder(),
lstm_models.HierarchicalLstmDecoder(
lstm_models.CategoricalLstmDecoder(),
level_lengths=[16, 16],
disable_autoregression=True)),
hparams=merge_hparams(
lstm_models.get_default_hparams(),
HParams(
batch_size=512,
max_seq_len=256,
z_size=512,
enc_rnn_size=[2048, 2048],
dec_rnn_size=[1024, 1024],
free_bits=256,
max_beta=0.2,
)),
note_sequence_augmenter=None,
data_converter=data.OneHotMelodyConverter(
skip_polyphony=False,
max_bars=100, # Truncate long melodies before slicing.
slice_bars=16,
steps_per_quarter=4),
train_examples_path=None,
eval_examples_path=None,
)
