model_upsample_network {torchaudio}

UpsampleNetwork

Description

Upscale the dimensions of a spectrogram. Pass the input through the UpsampleNetwork layer.

Usage

model_upsample_network(
  upsample_scales,
  n_res_block = 10,
  n_freq = 128,
  n_hidden = 128,
  n_output = 128,
  kernel_size = 5
)

Arguments

upsample_scales

the list of upsample scales.

n_res_block

the number of ResBlock in stack. (Default: 10)

n_freq

the number of bins in a spectrogram. (Default: 128)

n_hidden

the number of hidden dimensions of resblock. (Default: 128)

n_output

the number of output dimensions of melresnet. (Default: 128)

kernel_size

the number of kernel size in the first Conv1d layer. (Default: 5)

Details

forward param: specgram (Tensor): the input sequence to the UpsampleNetwork layer (n_batch, n_freq, n_time)

Value

Tensor shape: (n_batch, n_freq, (n_time - kernel_size + 1) * total_scale), (n_batch, n_output, (n_time - kernel_size + 1) * total_scale) where total_scale is the product of all elements in upsample_scales.

Examples

if(torch::torch_is_installed()) {
 upsamplenetwork = model_upsample_network(upsample_scales=c(4, 4, 16))
 input = torch::torch_rand (10, 128, 10)  # a random spectrogram
 output = upsamplenetwork (input)  # shape: (10, 1536, 128), (10, 1536, 128)
}