sadaco.dataman package

Subpackages

• sadaco.dataman.icbhi package
  • Submodules
  • sadaco.dataman.icbhi.demo module
  • sadaco.dataman.icbhi.dummy module
  • sadaco.dataman.icbhi.icbhi module
  • Module contents

Submodules

sadaco.dataman.base module

class sadaco.dataman.base.BaseDataset(*args, **kwds)

Bases: torch.utils.data.dataset.Dataset

Dataset Template

convert_wav(waveform)

Convert wav file to Mag+Phase matrix with STFT conversion. User can override this func to customize data format.

Parameters

waveform (torch.Tensor) – Input wav file. Required shape : [Batch, Length]

Returns

Tuple of mag, phase matrix

Return type

Tuple[Torch.Tensor]

recover_wav(mag, phase)

Inverse function of convert_wav. User should modify both of the functions when customizing.

Parameters

• mag (torch.Tensor) – Magnitude matrix from STFT.

• phase (torch.Tensor) – Phase matrix from STFT

Returns

Return type

_type_

load_datum(index)

static load_wav_from_path(path)

parse_label()

static db_to_linear(samples)

loudness_normalization(samples: torch.Tensor, target_db: float = 15.0, max_gain_db: float = 30.0)

sadaco.dataman.build_dataset module

sadaco.dataman.loader module

sadaco.dataman.sampler module

class sadaco.dataman.sampler.BalancedBatchSampler(*args, **kwds)

Bases: torch.utils.data.sampler.BatchSampler

BatchSampler - from a MNIST-like dataset, samples n_classes and within these classes samples n_samples. Returns batches of size n_classes * n_samples

Module contents