lbg

diffsptk.LBG

alias of LindeBuzoGrayAlgorithm

class diffsptk.LindeBuzoGrayAlgorithm(order, codebook_size, *, min_data_per_cluster=1, n_iter=100, eps=1e-05, perturb_factor=1e-05, init='mean', metric='none', batch_size=None, seed=None, verbose=False)

See this page for details. This module is not differentiable.

Parameters:

orderint >= 0

Order of vector, \(M\).

codebook_sizeint >= 1

Target codebook size, must be power of two, \(K\).

min_data_per_clusterint >= 1

Minimum number of data points in a cluster.

n_iterint >= 1

Number of iterations.

epsfloat >= 0

Convergence threshold.

perturb_factorfloat > 0

Perturbation factor.

init[‘none’, ‘mean’] or torch.Tensor [shape=(1~K, M+1)]

Initialization type.

metric[‘none, ‘aic’, ‘bic’]

Metric used as a reference for model selection.

batch_sizeint >= 1 or None

Batch size.

seedint or None

Random seed.

verbosebool or int

If 1, show distance at each iteration; if 2, show progress bar.

forward(x, return_indices=False)

Design a codebook.

Parameters:

xTensor [shape=(T, M+1)] or DataLoader

Input vectors or dataloder yielding input vectors.

return_indicesbool

If True, return indices.

Returns:

codebookTensor [shape=(K, M+1)]

Codebook.

indicesTensor [shape=(T,)] (optional)

Codebook indices.

distanceTensor [scalar]

Distance.

Examples

>>> x = diffsptk.nrand(10, 0)
>>> lbg = diffsptk.LBG(0, 2)
>>> codebook, indices, distance = lbg(x, return_indices=True)
>>> codebook
tensor([[-0.5277],
        [ 0.6747]])
>>> indices
tensor([0, 0, 0, 1, 0, 1, 1, 1, 1, 0])
>>> distance
tensor(0.2331)

transform(x)

Transform input vectors using the codebook.

Parameters:

xTensor [shape=(T, M+1)]

Input vectors.

Returns:

xqTensor [shape=(T, M+1)]

Quantized vectors.

indicesTensor [shape=(T,)]

Codebook indices.

Examples

>>> lbg = diffsptk.LBG(0, 2)
>>> torch.save(lbg.state_dict(), "lbg.pt")
>>> lbg.load_state_dict(torch.load("lbg.pt"))
>>> x = diffsptk.nrand(10, 0)
>>> xq, indices = lbg.transform(x)

See also

vq ivq gmm