seacj

`output_numpy = np.zeros(shape=[num_enrollment,40,128],dtype=np.float32) ` `model = np.zeros(shape=[40,128],dtype=np.float32) ` `outputs = net(inputs) # shape is (batch_size,128) ` `output_numpy[i] = outputs.cpu().data.numpy()` Here‘s the code in enrollment.py. I'm confused about the parameter 40....

In the paper for Similarity-Preserving loss. The normalization is before the operation of matrix Multiplication. Does the order matter the performance. ``` import torch org_f_s = torch.rand((64, 96)) org_f_t =...

你好 Gammatone滤波器组：性质、实现和应用写的非常好 我想请教关于FIR的实现。 你写道： ``` g = @(t, f0, erb) t .^ 3 .* exp(- 2 * pi * 1.019 * erb * t) .* cos(2 * pi * f0...

Could you provide the cohort files in runEcapaXvector

1

I noticed that the performance improve significantly in AS-norm for ECAPAXvector.(EER 1.506->1.140). Could you provide the cohort file you choose?

It seems that the motivation to design the jasper architecture is not clear. Most networks are constructed by stacking the basic module such as Resnet, MobileNet. The design of the...