Update MelResNet #751
Update MelResNet #751
Conversation
Codecov Report
@@ Coverage Diff @@
## master #751 +/- ##
==========================================
- Coverage 89.24% 89.22% -0.02%
==========================================
Files 32 32
Lines 2519 2515 -4
==========================================
- Hits 2248 2244 -4
Misses 271 271
Continue to review full report at Codecov.
|
jimchen90
changed the title
torchaudio/models/_wavernn.py
Outdated
| Florian Stimberg, Aaron van den Oord, Sander Dieleman, Koray Kavukcuoglu. arXiv:1802.08435, 2018. | ||
| r"""ResNet block based on "Deep Residual Learning for Image Recognition" | ||
|
|
||
| The paper link is https://arxiv.org/pdf/1512.03385.pdf. The input signal is spectrogram. |
There was a problem hiding this comment.
Updated.
I also updated this change in the _wavernn class in #735
test/test_models.py
Outdated
| n_batch = 2 | ||
| n_time = 200 | ||
| n_freq = 100 | ||
| n_output = 256 |
There was a problem hiding this comment.
why is the output shape changing?
There was a problem hiding this comment.
When I ran the test, I tried a different input value to double check if the test passes. Do you think I need to change it back to n_output = 128?
There was a problem hiding this comment.
Ok, thanks for clarifying. Let's change it back to what it was.
torchaudio/models/_wavernn.py
Outdated
| Examples:: | ||
| >>> resblock = _ResBlock(num_dims=128) | ||
| Examples | ||
| >>> resblock = _ResBlock() | ||
| >>> input = torch.rand(10, 128, 512) |
There was a problem hiding this comment.
in the example, let's make it clear we are discussing a spectrogram:
input = torch.rand(10, 128, 512) # a random spectrogram
output = resblock(input) # shape: (10, 128, 512)
test/test_models.py
Outdated
| @@ -36,18 +36,20 @@ def test_mfcc(self): | |||
| class TestMelResNet(common_utils.TorchaudioTestCase): | |||
|
|
|||
| def test_waveform(self): | |||
| """test the output dimensions after _MelResNet block. | |||
There was a problem hiding this comment.
nit: "Validate the output dimensions of a _MelResNet block."
torchaudio/models/_wavernn.py
Outdated
| Florian Stimberg, Aaron van den Oord, Sander Dieleman, Koray Kavukcuoglu. arXiv:1802.08435, 2018. | ||
| r"""ResNet block based on "Deep Residual Learning for Image Recognition" | ||
|
|
||
| The paper link is https://arxiv.org/pdf/1512.03385.pdf. The input signal is a spectrogram. |
There was a problem hiding this comment.
nit: after re-reading this, I would simply remove "The input signal is a spectrogram." since you are already describing the arguments below.
There was a problem hiding this comment.
I agree. I removed this sentence. Thanks.
Jlin tutorials quant
This is to update the variable names and docstring in #705
Stack:
Add MelResNet Block #705, #751Add Upsampling Block #724Add WaveRNN Model #735Add example pipeline with WaveRNN #749