DeepLearningExamples/PyTorch/SpeechSynthesis/FastPitch/waveglow/denoiser.py

# *****************************************************************************
#  Copyright (c) 2018, NVIDIA CORPORATION.  All rights reserved.
#
#  Redistribution and use in source and binary forms, with or without
#  modification, are permitted provided that the following conditions are met:
#      * Redistributions of source code must retain the above copyright
#        notice, this list of conditions and the following disclaimer.
#      * Redistributions in binary form must reproduce the above copyright
#        notice, this list of conditions and the following disclaimer in the
#        documentation and/or other materials provided with the distribution.
#      * Neither the name of the NVIDIA CORPORATION nor the
#        names of its contributors may be used to endorse or promote products
#        derived from this software without specific prior written permission.
#
#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
#  ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
#  WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
#  DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
#  DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
#  (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
#  LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
#  ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
#  (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
#  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#
# *****************************************************************************

import torch
from common.layers import STFT


class Denoiser(torch.nn.Module):
    """ Removes model bias from audio produced with waveglow """

    def __init__(self, waveglow, filter_length=1024, n_overlap=4,
                 win_length=1024, mode='zeros'):
        super(Denoiser, self).__init__()
        device = waveglow.upsample.weight.device
        dtype = waveglow.upsample.weight.dtype
        self.stft = STFT(filter_length=filter_length,
                         hop_length=int(filter_length/n_overlap),
                         win_length=win_length).to(device)
        if mode == 'zeros':
            mel_input = torch.zeros((1, 80, 88), dtype=dtype, device=device)
        elif mode == 'normal':
            mel_input = torch.randn((1, 80, 88), dtype=dtype, device=device)
        else:
            raise Exception("Mode {} if not supported".format(mode))

        with torch.no_grad():
            bias_audio = waveglow.infer(mel_input, sigma=0.0).float()
            bias_spec, _ = self.stft.transform(bias_audio)

        self.register_buffer('bias_spec', bias_spec[:, :, 0][:, :, None])

    def forward(self, audio, strength=0.1):
        audio_spec, audio_angles = self.stft.transform(audio)
        audio_spec_denoised = audio_spec - self.bias_spec * strength
        audio_spec_denoised = torch.clamp(audio_spec_denoised, 0.0)
        audio_denoised = self.stft.inverse(audio_spec_denoised, audio_angles)
        return audio_denoised