SpecAugment
SpecAugment copied to clipboard
DemisEom
A Implementation of SpecAugment with Tensorflow & Pytorch, introduced by Google Brain
---> 17 warped_masked_spectrogram = spec_augment_tensorflow.spec_augment(mel_spectrogram=mel_spectrogram) 18 print(warped_masked_spectrogram) /usr/local/lib/python3.7/dist-packages/specAugment/spec_augment_tensorflow.py in spec_augment(mel_spectrogram, time_warping_para, frequency_masking_para, time_masking_para, frequency_mask_num, time_mask_num) mtrand.pyx in numpy.random.mtrand.RandomState.randint() _bounded_integers.pyx in numpy.random._bounded_integers._rand_int64() can anyone help me how to solve this error
ModuleNotFoundError Traceback (most recent call last) [](https://localhost:8080/#) in () 32 import librosa, librosa.display 33 from audiomentations import Compose, AddGaussianNoise, FrequencyMask, TimeStretch, PitchShift, Shift ---> 34 from specAugment import spec_augment_tensorflow 35...
I am just curious what people use for Masking value. It seems this repo implements zero-value masking, zero value in a spectrogram does not necessarily mean "silence" if we operate...
https://github.com/seriousran/SpecAugment 1.Fixed typo in import for pytorch https://github.com/DemisEom/SpecAugment/pull/31 - https://github.com/seriousran/SpecAugment/commit/b1d218fcb8117639a66b86a75472bda12097e3c3 2. Update sparse_image_warp_pytorch.py https://github.com/DemisEom/SpecAugment/pull/33 - https://github.com/seriousran/SpecAugment/commit/1e5b59297ff27ddf0a63bc1cf0d7c308a65ab5a7 3. ImportError: Cannot load backend TkAgg requires tk - https://github.com/seriousran/SpecAugment/commit/39f849d279f2b06d1cfc2b8deb99452c7892e081 `pip install git+https://github.com/seriousran/SpecAugment.git` I'm...
when I try the audio file SpecAugment，something wrong happaned, do anyone have an idea to solve the following problem? Python 3.6.9 |Anaconda, Inc.| (default, Jul 30 2019, 19:07:31) [GCC 7.3.0]...
E.g, line 62 in spec_augment_tensorflow.py: ''' fbank_size = tf.shape(spectrogram) n, v = fbank_size[1], fbank_size[2] ''' And 'n' is used as the length of time, and 'v' is used as the...
There's sparse_image_warp_pytorch.py and sparse_image_warp_np.py, but they don't get used anywhere. In spec_augment_pytorch.py, sparse_image_warp gets never called and is still commented as TODO. The documentation should at least state that if...
When I use the spec_augment in my network,the training speed become very slowly.How to solve this problem?
For example, I have a tensor with a batch size of 2, e.g. of shape `torch.Size([2, 256, 512])`. Now, when I run the command through ``` spec_augment_pytorch.spec_augment(mel_spectrogram=mel_spectrogram) ``` I have...
