hparams.py

# -*- coding: utf-8 -*-
#/usr/bin/python2

# path
## local
data_path_base = './datasets'
logdir_path = './logdir'

## remote
# data_path_base = '/data/private/vc/datasets'
# logdir_path = '/data/private/vc/logdir'


class Default:
    # signal processing
    sr = 16000 # Sampling rate.
    frame_shift = 0.005  # seconds
    frame_length = 0.025  # seconds
    n_fft = 512

    hop_length = int(sr*frame_shift)  # 80 samples.  This is dependent on the frame_shift.
    win_length = int(sr*frame_length)  # 400 samples. This is dependent on the frame_length.

    preemphasis = 0.97
    n_mfcc = 40
    n_iter = 60 # Number of inversion iterations
    n_mels = 80
    duration = 2

    # mean_log_spec = -4.25
    # std_log_spec = 2.15
    # min_log_spec = -21.25
    # max_log_spec = 3.0

    # model
    hidden_units = 256  # alias = E
    num_banks = 16
    num_highway_blocks = 4
    norm_type = 'ins'  # a normalizer function. value: bn, ln, ins, or None
    t = 1.0  # temperature
    dropout_rate = 0.2

    # train
    batch_size = 32


class Train1:
    # path
    data_path = '{}/timit/TIMIT/TRAIN/*/*/*.wav'.format(data_path_base)

    # model
    hidden_units = 256  # alias = E
    num_banks = 16
    num_highway_blocks = 4
    norm_type = 'ins'  # a normalizer function. value: bn, ln, ins, or None
    t = 1.0  # temperature
    dropout_rate = 0.2

    # train
    batch_size = 32
    lr = 0.0003
    num_epochs = 1000
    save_per_epoch = 2


class Train2:
    # path
    data_path = '{}/kate/sense_and_sensibility_split/*.wav'.format(data_path_base)
    # data_path = '{}/arctic/slt/*.wav'.format(data_path_base)

    # model
    hidden_units = 512  # alias = E
    num_banks = 16
    num_highway_blocks = 8
    norm_type = 'ins'  # a normalizer function. value: bn, ln, ins, or None
    t = 1.0  # temperature
    dropout_rate = 0.2

    # train
    batch_size = 32
    lr = 0.0005
    num_epochs = 10000
    save_per_epoch = 50


class Test1:
    # path
    data_path = '{}/timit/TIMIT/TEST/*/*/*.wav'.format(data_path_base)

    # test
    batch_size = 32


class Test2:
    # test
    batch_size = 32


class Convert:
    # path
    data_path = '{}/arctic/bdl/*.wav'.format(data_path_base)

    # convert
    batch_size = 2
    emphasis_magnitude = 1.2