Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Make torch_stft_fb.py encoder onnxable #11

Merged
merged 1 commit into from
Mar 12, 2021
Merged

Make torch_stft_fb.py encoder onnxable #11

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 7 additions & 6 deletions asteroid_filterbanks/torch_stft_fb.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,10 +1,9 @@
import warnings
from typing import List
from typing import Tuple
import torch
import torch.nn.functional as F
from asteroid_filterbanks import STFTFB, Encoder, Decoder
from asteroid_filterbanks import STFTFB
from asteroid_filterbanks.scripting import script_if_tracing
import numpy as np


class TorchSTFTFB(STFTFB):
Expand Down Expand Up @@ -99,7 +98,7 @@ def pre_analysis(self, wav):
"""Centers the frames if `center` is True."""
if not self.center:
return wav
pad_shape = [self.kernel_size // 2, self.kernel_size // 2]
pad_shape = (self.kernel_size // 2, self.kernel_size // 2)
wav = pad_all_shapes(wav, pad_shape=pad_shape, mode=self.pad_mode)
return wav

Expand Down Expand Up @@ -178,12 +177,14 @@ def square_ola(window: torch.Tensor, kernel_size: int, stride: int, n_frame: int


@script_if_tracing
def pad_all_shapes(x: torch.Tensor, pad_shape: List[int], mode: str = "reflect") -> torch.Tensor:
def pad_all_shapes(
x: torch.Tensor, pad_shape: Tuple[int, int], mode: str = "reflect"
) -> torch.Tensor:
if x.ndim == 1:
return F.pad(x[None, None], pad=pad_shape, mode=mode).squeeze(0).squeeze(0)
if x.ndim == 2:
return F.pad(x[None], pad=pad_shape, mode=mode).squeeze(0)
if x.ndim == 3:
return F.pad(x, pad=pad_shape, mode=mode)
pad_shape = [pad_shape[0]] + [0 for _ in range(x.ndim - 1)]
pad_shape = (pad_shape[0],) + (0,) * (x.ndim - 1)
return F.pad(x, pad=pad_shape, mode=mode)