SpecAugment icon indicating copy to clipboard operation
SpecAugment copied to clipboard

A Implementation of SpecAugment with Tensorflow & Pytorch, introduced by Google Brain

Results 26 SpecAugment issues
Sort by recently updated
recently updated
newest added

I found it was kind of slow when do SpecAugment on the CPU; and I would like to know which device you run your SpecAugment program on, CPU or GPU?...

In the sparse_image_warp_np of sparse_image_warp_np.py file scipy.interpolate.spline is used but this function is not available in scipy 1.3.0. Using sparse_image_warp_np shows a call trace as follows. File "C:\Anaconda\lib\site-packages\specAugment\sparse_image_warp_np.py", line 121,...

In paper: "A random point along the horizontal line passing through the center of the image within the time steps (W, τ − W ) is to be warped either...

your code was error on 75th line in spec_augment_pytorch.py. log mel spectrogram wasn't changed by Step 1 : Time warping. the module was modified as follows: for i in range(v):...

Dear author, I have a question when look at Time Warp implementation: ```python def time_warp(spec, W=5): spec = spec.view(1, spec.shape[0], spec.shape[1]) num_rows = spec.shape[1] spec_len = spec.shape[2] y = num_rows...

It is not an issue, more like a note. image augmentation library Albumentations has transform XYSPec, which is a generalization of SpecAugment. Works on - uint8 and float32 images -...