Home

Welcome to the Pymuse Wiki!

I currently don't have that much time to write docs because I am working on code but there will be something here eventually

Usage

Train LSTM

from Models.lstm import ABCLSTM
from text_utils import get_vocab
# Will fill out all of the docs and readme stuff later

songs_joined = "\n\n".join(songs)
vocab = get_vocab(songs_joined)
char2idx = {u: i for i, u in enumerate(vocab)}
idx2char = np.array(vocab)
vectorized_songs = vectorize_string(char2idx, songs_joined)

model = ABCLSTM()
model.train_lstm(batch_size=32, char2idx=char2idx, idx2char=idx2char, learning_rate=1e-3, embedding_dim=256,
                 steps_per_epoch=5000, epochs=1, vocab_size=len(vocab), vectorized=vectorized_songs,
                 sequence_length=100, rnn_units=1024, output_path="Models\\train_lstm", output_name="name")

Train RNN

from Models.rnn import ABCGRURNN
from text_utils import get_vocab
# Will fill out all of the docs and readme stuff later
songs_joined = "\n\n".join(songs)
vocab = get_vocab(songs_joined)
char2idx = {u: i for i, u in enumerate(vocab)}
idx2char = np.array(vocab)
vectorized_songs = vectorize_string(char2idx, songs_joined)

model = ABCGRURNN()
model.train_lstm(batch_size=32, char2idx=char2idx, idx2char=idx2char, learning_rate=1e-3, embedding_dim=256,
                 steps_per_epoch=5000, epochs=1, vocab_size=len(vocab), vectorized=vectorized_songs,
                 sequence_length=100, rnn_units=1024, output_path="Models\\train_rnn", output_name="name")

Train SEQ2SEQ

from Models.seq2seq import ABCSEQ2SEQ
# Will fill out all of the docs and readme stuff later

Infer with ABCLSTM

from models.lstm import ABCLSTM
# Will fill out all of the docs and readme stuff later
model = ABCLSTM()
model.load_lstm_model(path="path\\to\\file", config_path="path\\to\\file", rnn_units=1024,
                    embedding_dim=256, vocab_size=len(vocab))
preds = model.predict_lstm_model(start_seed="X", generation_length=1000, format="midi", fp="out1")

Infer with ABCGRURNN

from models.rnn import ABCGRURNN
# Will fill out all of the docs and readme stuff later

model = ABCGRURNN()
model.load_rnn_model(path="path\\to\\file", config_path="path\\to\\file", rnn_units=1024,
                  embedding_dim=256, vocab_size=len(vocab))
preds = model.predict_rnn_model(start_seed="X", generation_length=1000, format="midi", fp="out1")