espnet2.gan_svs.avocodo.avocodo.MDCDConfig

source

class espnet2.gan_svs.avocodo.avocodo.MDCDConfig(h)

Bases: object

Configuration class for the Multi-band Discriminator (MDC).

This class holds the configuration parameters required for the Multi-band Discriminator, including PQMF parameters, filter sizes, kernel sizes, dilations, strides, band ranges, and segment size.

pqmf_params

Parameters for PQMF configuration.

• Type: List[int]

f_pqmf_params

Parameters for filtered PQMF configuration.

• Type: List[int]

filters

List of filter sizes for the discriminator.

• Type: List[List[int]]

kernel_sizes

List of kernel sizes for the convolutional layers.

• Type: List[List[int]]

dilations

List of dilations for the convolutional layers.

• Type: List[List[int]]

strides

List of strides for the convolutional layers.

• Type: List[List[int]]

band_ranges

List of ranges for each frequency band.

• Type: List[List[int]]

transpose

Indicates whether to transpose the input for each band.

• Type: List[bool]

segment_size

Size of segments to be processed.

• Type: int

• Parameters:h (Dict *[*str , Any ]) – A dictionary containing configuration parameters.

Examples

>>> config = MDCDConfig({
...     "pqmf_config": {"sbd": [16, 256, 0.03, 10.0],
...                     "fsbd": [64, 256, 0.1, 9.0]},
...     "sbd_filters": [[64, 128, 256], [32, 64, 128]],
...     "sbd_kernel_sizes": [[[3, 3, 3], [5, 5, 5]]],
...     "sbd_dilations": [[[1, 2, 3], [1, 2, 3]]],
...     "sbd_strides": [[1, 1, 1]],
...     "sbd_band_ranges": [[0, 16]],
...     "sbd_transpose": [False],
...     "segment_size": 8192
... })
>>> print(config.filters)
[[64, 128, 256], [32, 64, 128]]