blob: 753e76b3cbf45fecf11ddedae992446ef813110d [file] [log] [blame]
# Copyright 2023 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
import tensorflow as tf
from tflite_micro.python.tflite_micro.signal.utils import util
from tflite_micro.python.tflite_micro.signal.utils import wide_dynamic_func_lut_wrapper
gen_pcan_op = util.load_custom_op("pcan_op.so")
PCAN_SNR_BITS = 12
def _pcan_wrapper(pcan_fn, default_name):
"""Wrapper around gen_pcan.pcan*."""
def _pcan(input_tensor,
noise_estimate,
strength,
offset,
gain_bits,
smoothing_bits,
input_correction_bits,
name=default_name):
with tf.name_scope(name) as scope:
input_tensor = tf.convert_to_tensor(input_tensor, dtype=tf.uint32)
noise_estimate = tf.convert_to_tensor(noise_estimate, dtype=tf.uint32)
input_bits = smoothing_bits - input_correction_bits
snr_shift = gain_bits - input_correction_bits - PCAN_SNR_BITS
if snr_shift < 1:
raise ValueError("SNR shift must be non-negative: %d" % snr_shift)
lut = wide_dynamic_func_lut_wrapper.wide_dynamic_func_lut(
strength, offset, input_bits, gain_bits)
lut_tensor = tf.convert_to_tensor(lut, dtype=tf.int16)
dim_list = input_tensor.shape.as_list()
if len(dim_list) != 1:
raise ValueError("Input tensor must have a rank of 1")
dim_list = noise_estimate.shape.as_list()
if len(dim_list) != 1:
raise ValueError("Noise estimate must have a rank of 1")
snr_shift = 6
return pcan_fn(input_tensor,
noise_estimate,
lut_tensor,
snr_shift=snr_shift,
name=scope)
return _pcan
# TODO(b/286250473): change back name after name clash resolved
pcan = _pcan_wrapper(gen_pcan_op.signal_pcan, "signal_pcan")
tf.no_gradient("pcan")