plot audio in SummaryWriter

plot_audio.py

#!/usr/bin/python
# -*- coding: utf-8 -*-
# Copyright 2022 Lucky Wong
#
# 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 matplotlib.pyplot as plt
import numpy as np

tgt = target[0].cpu().detach().numpy()
noisy = probe[0].cpu().detach().numpy()
pred = prediction[0].cpu().detach().numpy()

fig, axs = plt.subplots(nrows=3, ncols=2, figsize=(10, 20))

for idx, info in enumerate([(tgt, "tgt"), (noisy, "noisy"), (pred, "pred")]):
    # wave
    y, title = info
    axs[idx, 0].plot(y)
    axs[idx, 0].set_title(f'Waveform of {title}')
    axs[idx, 0].set_xlabel('Time (samples)')
    axs[idx, 0].set_ylabel('Amplitude')

    # specgram
    axs[idx, 1].specgram(y, NFFT=512, Fs=16000,
                         cmap='viridis', mode='magnitude')
    axs[idx, 1].set_title(f'Log-frequency power spectrogram of {title}')
    axs[idx, 1].set_xlabel('Time (seconds)')
    axs[idx, 1].set_ylabel('Frequency (Hz)')

fig.canvas.draw()
w, h = fig.canvas.get_width_height()
image = np.frombuffer(fig.canvas.tostring_rgb(),
                      dtype='uint8').reshape((h, w, 3))
image = torch.from_numpy(image).permute(2, 0, 1)
self.writer.add_image('Audio', image, global_step=0)