File size: 12,701 Bytes
9c6594c |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 |
# Copyright The Lightning team.
#
# 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.
from collections.abc import Sequence
from typing import Any, Optional, Union
from torch import Tensor, tensor
from torchmetrics.functional.audio.snr import (
complex_scale_invariant_signal_noise_ratio,
scale_invariant_signal_noise_ratio,
signal_noise_ratio,
)
from torchmetrics.metric import Metric
from torchmetrics.utilities.imports import _MATPLOTLIB_AVAILABLE
from torchmetrics.utilities.plot import _AX_TYPE, _PLOT_OUT_TYPE
if not _MATPLOTLIB_AVAILABLE:
__doctest_skip__ = [
"SignalNoiseRatio.plot",
"ScaleInvariantSignalNoiseRatio.plot",
"ComplexScaleInvariantSignalNoiseRatio.plot",
]
class SignalNoiseRatio(Metric):
r"""Calculate `Signal-to-noise ratio`_ (SNR_) meric for evaluating quality of audio.
.. math::
\text{SNR} = \frac{P_{signal}}{P_{noise}}
where :math:`P` denotes the power of each signal. The SNR metric compares the level of the desired signal to
the level of background noise. Therefore, a high value of SNR means that the audio is clear.
As input to `forward` and `update` the metric accepts the following input
- ``preds`` (:class:`~torch.Tensor`): float tensor with shape ``(...,time)``
- ``target`` (:class:`~torch.Tensor`): float tensor with shape ``(...,time)``
As output of `forward` and `compute` the metric returns the following output
- ``snr`` (:class:`~torch.Tensor`): float scalar tensor with average SNR value over samples
Args:
zero_mean: if to zero mean target and preds or not
kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info.
Raises:
TypeError:
if target and preds have a different shape
Example:
>>> from torch import tensor
>>> from torchmetrics.audio import SignalNoiseRatio
>>> target = tensor([3.0, -0.5, 2.0, 7.0])
>>> preds = tensor([2.5, 0.0, 2.0, 8.0])
>>> snr = SignalNoiseRatio()
>>> snr(preds, target)
tensor(16.1805)
"""
full_state_update: bool = False
is_differentiable: bool = True
higher_is_better: bool = True
sum_snr: Tensor
total: Tensor
plot_lower_bound: Optional[float] = None
plot_upper_bound: Optional[float] = None
def __init__(
self,
zero_mean: bool = False,
**kwargs: Any,
) -> None:
super().__init__(**kwargs)
self.zero_mean = zero_mean
self.add_state("sum_snr", default=tensor(0.0), dist_reduce_fx="sum")
self.add_state("total", default=tensor(0), dist_reduce_fx="sum")
def update(self, preds: Tensor, target: Tensor) -> None:
"""Update state with predictions and targets."""
snr_batch = signal_noise_ratio(preds=preds, target=target, zero_mean=self.zero_mean)
self.sum_snr += snr_batch.sum()
self.total += snr_batch.numel()
def compute(self) -> Tensor:
"""Compute metric."""
return self.sum_snr / self.total
def plot(
self, val: Optional[Union[Tensor, Sequence[Tensor]]] = None, ax: Optional[_AX_TYPE] = None
) -> _PLOT_OUT_TYPE:
"""Plot a single or multiple values from the metric.
Args:
val: Either a single result from calling `metric.forward` or `metric.compute` or a list of these results.
If no value is provided, will automatically call `metric.compute` and plot that result.
ax: An matplotlib axis object. If provided will add plot to that axis
Returns:
Figure and Axes object
Raises:
ModuleNotFoundError:
If `matplotlib` is not installed
.. plot::
:scale: 75
>>> # Example plotting a single value
>>> import torch
>>> from torchmetrics.audio import SignalNoiseRatio
>>> metric = SignalNoiseRatio()
>>> metric.update(torch.rand(4), torch.rand(4))
>>> fig_, ax_ = metric.plot()
.. plot::
:scale: 75
>>> # Example plotting multiple values
>>> import torch
>>> from torchmetrics.audio import SignalNoiseRatio
>>> metric = SignalNoiseRatio()
>>> values = [ ]
>>> for _ in range(10):
... values.append(metric(torch.rand(4), torch.rand(4)))
>>> fig_, ax_ = metric.plot(values)
"""
return self._plot(val, ax)
class ScaleInvariantSignalNoiseRatio(Metric):
"""Calculate `Scale-invariant signal-to-noise ratio`_ (SI-SNR) metric for evaluating quality of audio.
As input to `forward` and `update` the metric accepts the following input
- ``preds`` (:class:`~torch.Tensor`): float tensor with shape ``(...,time)``
- ``target`` (:class:`~torch.Tensor`): float tensor with shape ``(...,time)``
As output of `forward` and `compute` the metric returns the following output
- ``si_snr`` (:class:`~torch.Tensor`): float scalar tensor with average SI-SNR value over samples
Args:
kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info.
Raises:
TypeError:
if target and preds have a different shape
Example:
>>> import torch
>>> from torch import tensor
>>> from torchmetrics.audio import ScaleInvariantSignalNoiseRatio
>>> target = tensor([3.0, -0.5, 2.0, 7.0])
>>> preds = tensor([2.5, 0.0, 2.0, 8.0])
>>> si_snr = ScaleInvariantSignalNoiseRatio()
>>> si_snr(preds, target)
tensor(15.0918)
"""
is_differentiable = True
sum_si_snr: Tensor
total: Tensor
higher_is_better = True
plot_lower_bound: Optional[float] = None
plot_upper_bound: Optional[float] = None
def __init__(
self,
**kwargs: Any,
) -> None:
super().__init__(**kwargs)
self.add_state("sum_si_snr", default=tensor(0.0), dist_reduce_fx="sum")
self.add_state("total", default=tensor(0), dist_reduce_fx="sum")
def update(self, preds: Tensor, target: Tensor) -> None:
"""Update state with predictions and targets."""
si_snr_batch = scale_invariant_signal_noise_ratio(preds=preds, target=target)
self.sum_si_snr += si_snr_batch.sum()
self.total += si_snr_batch.numel()
def compute(self) -> Tensor:
"""Compute metric."""
return self.sum_si_snr / self.total
def plot(self, val: Union[Tensor, Sequence[Tensor], None] = None, ax: Optional[_AX_TYPE] = None) -> _PLOT_OUT_TYPE:
"""Plot a single or multiple values from the metric.
Args:
val: Either a single result from calling `metric.forward` or `metric.compute` or a list of these results.
If no value is provided, will automatically call `metric.compute` and plot that result.
ax: An matplotlib axis object. If provided will add plot to that axis
Returns:
Figure and Axes object
Raises:
ModuleNotFoundError:
If `matplotlib` is not installed
.. plot::
:scale: 75
>>> # Example plotting a single value
>>> import torch
>>> from torchmetrics.audio import ScaleInvariantSignalNoiseRatio
>>> metric = ScaleInvariantSignalNoiseRatio()
>>> metric.update(torch.rand(4), torch.rand(4))
>>> fig_, ax_ = metric.plot()
.. plot::
:scale: 75
>>> # Example plotting multiple values
>>> import torch
>>> from torchmetrics.audio import ScaleInvariantSignalNoiseRatio
>>> metric = ScaleInvariantSignalNoiseRatio()
>>> values = [ ]
>>> for _ in range(10):
... values.append(metric(torch.rand(4), torch.rand(4)))
>>> fig_, ax_ = metric.plot(values)
"""
return self._plot(val, ax)
class ComplexScaleInvariantSignalNoiseRatio(Metric):
"""Calculate `Complex scale-invariant signal-to-noise ratio`_ (C-SI-SNR) metric for evaluating quality of audio.
As input to `forward` and `update` the metric accepts the following input
- ``preds`` (:class:`~torch.Tensor`): real float tensor with shape ``(...,frequency,time,2)`` or complex float
tensor with shape ``(..., frequency,time)``
- ``target`` (:class:`~torch.Tensor`): real float tensor with shape ``(...,frequency,time,2)`` or complex float
tensor with shape ``(..., frequency,time)``
As output of `forward` and `compute` the metric returns the following output
- ``c_si_snr`` (:class:`~torch.Tensor`): float scalar tensor with average C-SI-SNR value over samples
Args:
zero_mean: if to zero mean target and preds or not
kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info.
Raises:
ValueError:
If ``zero_mean`` is not an bool
TypeError:
If ``preds`` is not the shape (..., frequency, time, 2) (after being converted to real if it is complex).
If ``preds`` and ``target`` does not have the same shape.
Example:
>>> from torch import randn
>>> from torchmetrics.audio import ComplexScaleInvariantSignalNoiseRatio
>>> preds = randn((1,257,100,2))
>>> target = randn((1,257,100,2))
>>> c_si_snr = ComplexScaleInvariantSignalNoiseRatio()
>>> c_si_snr(preds, target)
tensor(-38.8832)
"""
is_differentiable = True
ci_snr_sum: Tensor
num: Tensor
higher_is_better = True
plot_lower_bound: Optional[float] = None
plot_upper_bound: Optional[float] = None
def __init__(
self,
zero_mean: bool = False,
**kwargs: Any,
) -> None:
super().__init__(**kwargs)
if not isinstance(zero_mean, bool):
raise ValueError(f"Expected argument `zero_mean` to be an bool, but got {zero_mean}")
self.zero_mean = zero_mean
self.add_state("ci_snr_sum", default=tensor(0.0), dist_reduce_fx="sum")
self.add_state("num", default=tensor(0), dist_reduce_fx="sum")
def update(self, preds: Tensor, target: Tensor) -> None:
"""Update state with predictions and targets."""
v = complex_scale_invariant_signal_noise_ratio(preds=preds, target=target, zero_mean=self.zero_mean)
self.ci_snr_sum += v.sum()
self.num += v.numel()
def compute(self) -> Tensor:
"""Compute metric."""
return self.ci_snr_sum / self.num
def plot(self, val: Union[Tensor, Sequence[Tensor], None] = None, ax: Optional[_AX_TYPE] = None) -> _PLOT_OUT_TYPE:
"""Plot a single or multiple values from the metric.
Args:
val: Either a single result from calling `metric.forward` or `metric.compute` or a list of these results.
If no value is provided, will automatically call `metric.compute` and plot that result.
ax: An matplotlib axis object. If provided will add plot to that axis
Returns:
Figure and Axes object
Raises:
ModuleNotFoundError:
If `matplotlib` is not installed
.. plot::
:scale: 75
>>> # Example plotting a single value
>>> import torch
>>> from torchmetrics.audio import ComplexScaleInvariantSignalNoiseRatio
>>> metric = ComplexScaleInvariantSignalNoiseRatio()
>>> metric.update(torch.rand(1,257,100,2), torch.rand(1,257,100,2))
>>> fig_, ax_ = metric.plot()
.. plot::
:scale: 75
>>> # Example plotting multiple values
>>> import torch
>>> from torchmetrics.audio import ComplexScaleInvariantSignalNoiseRatio
>>> metric = ComplexScaleInvariantSignalNoiseRatio()
>>> values = [ ]
>>> for _ in range(10):
... values.append(metric(torch.rand(1,257,100,2), torch.rand(1,257,100,2)))
>>> fig_, ax_ = metric.plot(values)
"""
return self._plot(val, ax)
|