experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Quick return with simpler reshape op if the frame hop and frame length are statically known and their values are equal (i.e. no overlap). #tf-signal

Browse Source 
PiperOrigin-RevId: 208002136
This commit is contained in:
A. Unique TensorFlower 2018-08-08 23:50:57 -07:00 committed by TensorFlower Gardener
parent 4ff625bc58
commit f2494ede82
1 changed files with 16 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
tensorflow/contrib/signal/python/ops/reconstruction_ops.py
View File

@ -90,22 +90,28 @@ def overlap_and_add(signal, frame_step, name=None):
raise ValueError("frame_step must be an integer. Got %s" %
frame_step.dtype)
# If frame_length and frame_step are known at graph construction time, check
# frame_step is less than or equal to frame_length.
frame_step_static = tensor_util.constant_value(frame_step)
if (frame_step_static is not None and signal.shape.ndims is not None and
signal.shape[-1].value is not None and
frame_step_static > signal.shape[-1].value):
raise ValueError(
"frame_step (%d) must be less than or equal to frame_length (%d)" % (
frame_step_static, signal.shape[-1].value))
signal_shape = array_ops.shape(signal)
# All dimensions that are not part of the overlap-and-add. Can be empty for
# rank 2 inputs.
outer_dimensions = signal_shape[:-2]
# If frame_length and frame_step are known at graph construction time, check
# frame_step is less than or equal to frame_length.
frame_step_static = tensor_util.constant_value(frame_step)
if (frame_step_static is not None and signal.shape.ndims is not None and
signal.shape[-1].value is not None):
if frame_step_static > signal.shape[-1].value:
raise ValueError(
"frame_step (%d) must be less than or equal to "
"frame_length (%d)" % (
frame_step_static, signal.shape[-1].value))
# If frame_length is equal to frame_step, there's no overlap so just
# reshape the tensor.
if frame_step_static == signal.shape[-1].value:
return array_ops.reshape(signal, array_ops.concat(
[outer_dimensions, [-1]], 0))
signal_rank = array_ops.rank(signal)
frames = signal_shape[-2]
frame_length = signal_shape[-1]