File size: 10,693 Bytes
2777fde |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 |
"""
Implementation of objective functions used in the task 'End-to-end Remastering System'
"""
import numpy as np
import torch
import torch.nn.functional as F
import torch.nn as nn
import os
import sys
currentdir = os.path.dirname(os.path.realpath(__file__))
sys.path.append(os.path.dirname(currentdir))
from modules.training_utils import *
from modules.front_back_end import *
'''
Normalized Temperature-scaled Cross Entropy (NT-Xent) Loss
below source code (class NT_Xent) is a replication from the github repository - https://github.com/Spijkervet/SimCLR
the original implementation can be found here: https://github.com/Spijkervet/SimCLR/blob/master/simclr/modules/nt_xent.py
'''
class NT_Xent(nn.Module):
def __init__(self, batch_size, temperature, world_size):
super(NT_Xent, self).__init__()
self.batch_size = batch_size
self.temperature = temperature
self.world_size = world_size
self.mask = self.mask_correlated_samples(batch_size, world_size)
self.criterion = nn.CrossEntropyLoss(reduction="sum")
self.similarity_f = nn.CosineSimilarity(dim=2)
def mask_correlated_samples(self, batch_size, world_size):
N = 2 * batch_size * world_size
mask = torch.ones((N, N), dtype=bool)
mask = mask.fill_diagonal_(0)
for i in range(batch_size * world_size):
mask[i, batch_size + i] = 0
mask[batch_size + i, i] = 0
# mask[i, batch_size * world_size + i] = 0
# mask[batch_size * world_size + i, i] = 0
return mask
def forward(self, z_i, z_j):
"""
We do not sample negative examples explicitly.
Instead, given a positive pair, similar to (Chen et al., 2017), we treat the other 2(N − 1) augmented examples within a minibatch as negative examples.
"""
N = 2 * self.batch_size * self.world_size
z = torch.cat((z_i, z_j), dim=0)
# combine embeddings from all GPUs
if self.world_size > 1:
z = torch.cat(GatherLayer.apply(z), dim=0)
sim = self.similarity_f(z.unsqueeze(1), z.unsqueeze(0)) / self.temperature
sim_i_j = torch.diag(sim, self.batch_size * self.world_size)
sim_j_i = torch.diag(sim, -self.batch_size * self.world_size)
# We have 2N samples, but with Distributed training every GPU gets N examples too, resulting in: 2xNxN
positive_samples = torch.cat((sim_i_j, sim_j_i), dim=0).reshape(N, 1)
negative_samples = sim[self.mask].reshape(N, -1)
labels = torch.zeros(N).to(positive_samples.device).long()
logits = torch.cat((positive_samples, negative_samples), dim=1)
loss = self.criterion(logits, labels)
loss /= N
return loss
# Root Mean Squared Loss
# penalizes the volume factor with non-linearlity
class RMSLoss(nn.Module):
def __init__(self, reduce, loss_type="l2"):
super(RMSLoss, self).__init__()
self.weight_factor = 100.
if loss_type=="l2":
self.loss = nn.MSELoss(reduce=None)
def forward(self, est_targets, targets):
est_targets = est_targets.reshape(est_targets.shape[0]*est_targets.shape[1], est_targets.shape[2])
targets = targets.reshape(targets.shape[0]*targets.shape[1], targets.shape[2])
normalized_est = torch.sqrt(torch.mean(est_targets**2, dim=-1))
normalized_tgt = torch.sqrt(torch.mean(targets**2, dim=-1))
weight = torch.clamp(torch.abs(normalized_tgt-normalized_est), min=1/self.weight_factor) * self.weight_factor
return torch.mean(weight**1.5 * self.loss(normalized_est, normalized_tgt))
# Multi-Scale Spectral Loss proposed at the paper "DDSP: DIFFERENTIABLE DIGITAL SIGNAL PROCESSING" (https://arxiv.org/abs/2001.04643)
# we extend this loss by applying it to mid/side channels
class MultiScale_Spectral_Loss_MidSide_DDSP(nn.Module):
def __init__(self, mode='midside', \
reduce=True, \
n_filters=None, \
windows_size=None, \
hops_size=None, \
window="hann", \
eps=1e-7, \
device=torch.device("cpu")):
super(MultiScale_Spectral_Loss_MidSide_DDSP, self).__init__()
self.mode = mode
self.eps = eps
self.mid_weight = 0.5 # value in the range of 0.0 ~ 1.0
self.logmag_weight = 0.1
if n_filters is None:
n_filters = [4096, 2048, 1024, 512]
# n_filters = [4096]
if windows_size is None:
windows_size = [4096, 2048, 1024, 512]
# windows_size = [4096]
if hops_size is None:
hops_size = [1024, 512, 256, 128]
# hops_size = [1024]
self.multiscales = []
for i in range(len(windows_size)):
cur_scale = {'window_size' : float(windows_size[i])}
if self.mode=='midside':
cur_scale['front_end'] = FrontEnd(channel='mono', \
n_fft=n_filters[i], \
hop_length=hops_size[i], \
win_length=windows_size[i], \
window=window, \
device=device)
elif self.mode=='ori':
cur_scale['front_end'] = FrontEnd(channel='stereo', \
n_fft=n_filters[i], \
hop_length=hops_size[i], \
win_length=windows_size[i], \
window=window, \
device=device)
self.multiscales.append(cur_scale)
self.objective_l1 = nn.L1Loss(reduce=reduce)
self.objective_l2 = nn.MSELoss(reduce=reduce)
def forward(self, est_targets, targets):
if self.mode=='midside':
return self.forward_midside(est_targets, targets)
elif self.mode=='ori':
return self.forward_ori(est_targets, targets)
def forward_ori(self, est_targets, targets):
total_loss = 0.0
total_mag_loss = 0.0
total_logmag_loss = 0.0
for cur_scale in self.multiscales:
est_mag = cur_scale['front_end'](est_targets, mode=["mag"])
tgt_mag = cur_scale['front_end'](targets, mode=["mag"])
mag_loss = self.magnitude_loss(est_mag, tgt_mag)
logmag_loss = self.log_magnitude_loss(est_mag, tgt_mag)
# cur_loss = mag_loss + logmag_loss
# total_loss += cur_loss
total_mag_loss += mag_loss
total_logmag_loss += logmag_loss
# return total_loss
# print(f"ori - mag : {total_mag_loss}\tlog mag : {total_logmag_loss}")
return (1-self.logmag_weight)*total_mag_loss + \
(self.logmag_weight)*total_logmag_loss
def forward_midside(self, est_targets, targets):
est_mid, est_side = self.to_mid_side(est_targets)
tgt_mid, tgt_side = self.to_mid_side(targets)
total_loss = 0.0
total_mag_loss = 0.0
total_logmag_loss = 0.0
for cur_scale in self.multiscales:
est_mid_mag = cur_scale['front_end'](est_mid, mode=["mag"])
est_side_mag = cur_scale['front_end'](est_side, mode=["mag"])
tgt_mid_mag = cur_scale['front_end'](tgt_mid, mode=["mag"])
tgt_side_mag = cur_scale['front_end'](tgt_side, mode=["mag"])
mag_loss = self.mid_weight*self.magnitude_loss(est_mid_mag, tgt_mid_mag) + \
(1-self.mid_weight)*self.magnitude_loss(est_side_mag, tgt_side_mag)
logmag_loss = self.mid_weight*self.log_magnitude_loss(est_mid_mag, tgt_mid_mag) + \
(1-self.mid_weight)*self.log_magnitude_loss(est_side_mag, tgt_side_mag)
# cur_loss = mag_loss + logmag_loss
# total_loss += cur_loss
total_mag_loss += mag_loss
total_logmag_loss += logmag_loss
# return total_loss
# print(f"midside - mag : {total_mag_loss}\tlog mag : {total_logmag_loss}")
return (1-self.logmag_weight)*total_mag_loss + \
(self.logmag_weight)*total_logmag_loss
def to_mid_side(self, stereo_in):
mid = stereo_in[:,0] + stereo_in[:,1]
side = stereo_in[:,0] - stereo_in[:,1]
return mid, side
def magnitude_loss(self, est_mag_spec, tgt_mag_spec):
return torch.norm(self.objective_l1(est_mag_spec, tgt_mag_spec))
def log_magnitude_loss(self, est_mag_spec, tgt_mag_spec):
est_log_mag_spec = torch.log10(est_mag_spec+self.eps)
tgt_log_mag_spec = torch.log10(tgt_mag_spec+self.eps)
return self.objective_l2(est_log_mag_spec, tgt_log_mag_spec)
# hinge loss for discriminator
def dis_hinge(dis_fake, dis_real):
return torch.mean(torch.relu(1. - dis_real)) + torch.mean(torch.relu(1. + dis_fake))
# hinge loss for generator
def gen_hinge(dis_fake, dis_real=None):
return -torch.mean(dis_fake)
# DirectCLR's implementation of infoNCE loss
def infoNCE(nn, p, temperature=0.1):
nn = torch.nn.functional.normalize(nn, dim=1)
p = torch.nn.functional.normalize(p, dim=1)
nn = gather_from_all(nn)
p = gather_from_all(p)
logits = nn @ p.T
logits /= temperature
n = p.shape[0]
labels = torch.arange(0, n, dtype=torch.long).cuda()
loss = torch.nn.functional.cross_entropy(logits, labels)
return loss
# Class of available loss functions
class Loss:
def __init__(self, args, reduce=True):
device = torch.device("cpu")
if torch.cuda.is_available():
device = torch.device(f"cuda:{args.gpu}")
self.l1 = nn.L1Loss(reduce=reduce)
self.mse = nn.MSELoss(reduce=reduce)
self.ce = nn.CrossEntropyLoss()
self.triplet = nn.TripletMarginLoss(margin=1., p=2)
# self.ntxent = NT_Xent(args.train_batch*2, args.temperature, world_size=len(args.using_gpu.split(",")))
self.ntxent = NT_Xent(args.batch_size_total*(args.num_strong_negatives+1), args.temperature, world_size=1)
self.multi_scale_spectral_midside = MultiScale_Spectral_Loss_MidSide_DDSP(mode='midside', eps=args.eps, device=device)
self.multi_scale_spectral_ori = MultiScale_Spectral_Loss_MidSide_DDSP(mode='ori', eps=args.eps, device=device)
self.gain = RMSLoss(reduce=reduce)
self.infonce = infoNCE
|