training_sfe2.py

# -*- coding: utf-8 -*-
"""Training script for Sinusoidal Frequency Estimation by Gradient Descent.

Copyright (C) 2022 by Akira TAMAMORI
Copyright (C) 2022 by @xiao_ming

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
"""
import os

import librosa
import matplotlib.pyplot as plt
import numpy as np
import torch
from hydra import compose, initialize
from progressbar import progressbar
from torch import nn


class SFE(nn.Module):
    """Sinusoidal Frequency Estimation (SFE) class."""

    def __init__(self, cfg, device):
        """Initialize class."""
        super().__init__()
        n_sines = cfg.model.n_sines
        init_freq = torch.rand(n_sines, 1) * torch.pi
        init_phase = torch.rand(n_sines, 1) * torch.pi
        self._omega = nn.Parameter(torch.exp(init_freq * 1.0j))
        self._phi = nn.Parameter(torch.exp(init_phase * 1.0j))
        self.n_sines = n_sines
        self.arr_n = torch.arange(cfg.feature.frame_length).to(device)
        self.criterion = nn.MSELoss()
        self.config = cfg

    def forward(self):
        """Perform forward propagation."""
        mag = self._omega.abs().pow(self.arr_n) * self._phi.abs()
        sinusoid = torch.cos(self._omega.angle() * self.arr_n + self._phi.angle())
        estimated = mag * sinusoid
        estimated = torch.sum(estimated, dim=0)
        return estimated

    def get_loss(self, target):
        """Return MSE loss."""
        estimated_sinusoid = self.forward()
        loss = self.criterion(estimated_sinusoid, target)
        return loss

    def get_f0(self):
        """Return fundamental frequency in Hz."""
        freq = self._omega[:].angle() * self.config.feature.sample_rate / (2 * np.pi)
        freq = freq.to("cpu").detach().numpy().copy()  # tensor to numpy
        return np.min(np.abs(freq))

    @property
    def omega(self):
        """Get omega."""
        return self._omega

    @property
    def phi(self):
        """Get phi."""
        return self._phi


def get_power_max_frame(audio, frame_length, hop_length):
    """Extract the frame with maximum power."""
    frames = librosa.util.frame(
        audio, frame_length=frame_length, hop_length=hop_length
    ).T
    max_ind = np.argmax(np.sum(frames * frames, axis=1))
    max_frame = frames[max_ind, :]
    max_frame = max_frame / np.abs(np.max(max_frame))  # normalise within [-1, 1]
    return max_frame


def plot_waveform(audio, estimated, model, train_loss, cfg):
    """Extract the frame with maximum power."""
    sample_rate = cfg.feature.sample_rate
    fig = plt.figure(figsize=(cfg.demo.figsize.width, cfg.demo.figsize.height))
    time = np.arange(len(audio)) / sample_rate
    axes = fig.add_subplot(1, 1, 1)
    axes.plot(time, audio, label="original")
    axes.plot(time, estimated, label="estimated")
    axes.set_xlabel("Time (sec)", fontsize=cfg.demo.fontsize)
    axes.set_ylabel("Amplitude", fontsize=cfg.demo.fontsize)
    axes.set_title(f"Loss = {train_loss:.6f}", fontsize=cfg.demo.fontsize)
    axes.tick_params(axis="x", labelsize=cfg.demo.fontsize)
    axes.tick_params(axis="y", labelsize=cfg.demo.fontsize)
    axes.legend(fontsize=14)
    plt.savefig(cfg.demo.out_image)
    plt.show()


def training_loop(target, model, cfg):
    """Perform training loop."""
    optimizer = torch.optim.Adam(
        [model.omega, model.phi], lr=cfg.training.learning_rate
    )
    train_loss = 0.0
    for _ in progressbar(range(1, cfg.training.n_epochs)):
        optimizer.zero_grad()
        loss = model.get_loss(target)
        loss.backward()
        optimizer.step()
        train_loss += loss.item()

    train_loss /= cfg.training.n_epochs
    print(f"Loss: {train_loss}")

    return train_loss


def main(wav_file, cfg):
    """Perform model training."""
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

    # load target waveform
    wav_file = os.path.join(os.path.join(cfg.SFE.root_dir, cfg.SFE.data_dir), wav_file)
    audio, _ = librosa.load(wav_file, sr=cfg.feature.sample_rate)
    max_frame = get_power_max_frame(
        audio, cfg.feature.frame_length, cfg.feature.hop_length
    )
    target = torch.from_numpy(max_frame.astype(np.float32)).clone()  # numpy to tensor
    target = target.to(device)

    # instantiate model for training
    model = SFE(cfg, device).to(device)

    # perform training loop
    train_loss = training_loop(target, model, cfg)

    # plot estimated sinusoid wave
    estimated = model.forward()
    estimated = estimated.to("cpu").detach().numpy().copy()
    plot_waveform(max_frame, estimated, model, train_loss, cfg)


if __name__ == "__main__":
    with initialize(version_base=None, config_path="."):
        config = compose(config_name="config")
    main(config.demo.in_wav, config)